Synchronizing local compiler cache.
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- .gitattributes +21 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/Qwen/Qwen3-30B-A3B-Instruct-2507/2e075469bb7e246e19de.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/63936687f0615be0c4b5.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/67d11d4985c28ac7d5ff.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/cd03fa772b07fbb454c0.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/e5eddc28add59e2ee9d6.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/tiny-random/qwen3-moe/21c419fc8bcc3104cc37.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/tiny-random/qwen3-moe/e14cd546176a864ee18b.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/tiny-random/qwen3-moe/e99ce6fcf5be305e9424.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/yujiepan/qwen3-moe-tiny-random/1cacc2cc42865a8b000d.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/yujiepan/qwen3-moe-tiny-random/619f2f5356bdf0f59205.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/yujiepan/qwen3-moe-tiny-random/67bc8c12b9b1221b38ad.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/0f44022ffda5d90427b3.json +72 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/1471022e22b5b8b3de3b.json +72 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/c9c911d79dbddb873dcf.json +72 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/ibm-granite/granite-3.1-2b-instruct/b66dbd045f3e73eb7427.json +72 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/20b7299fc104cc69ce0f.json +76 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/502c5523ff7585cd3287.json +76 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/9c27c3f4a94313192c63.json +76 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/f0cd6ae3a57125c25463.json +76 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/unsloth/Llama-3.2-1B-Instruct/d4cdf18983e9784091ec.json +77 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/mixtral/dacorvo/Mixtral-tiny/34134d6beddf69d6e6c0.json +72 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/mixtral/dacorvo/Mixtral-tiny/4625dfce3a6ec89cfa42.json +72 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/mixtral/dacorvo/Mixtral-tiny/f790967cf96542a801bc.json +72 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/microsoft/Phi-3-mini-4k-instruct/c3b4f930a6c51ad40028.json +76 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/yujiepan/phi-4-tiny-random/63d7815d3de40f55a3ae.json +73 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/yujiepan/phi-4-tiny-random/892d9f2862662e5407c6.json +73 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/yujiepan/phi-4-tiny-random/e97f2d9bfc450d08ef3b.json +73 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/Qwen/Qwen2.5-0.5B/00b909b9addfb3e82c75.json +96 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/Qwen/Qwen2.5-0.5B/8a714805c0774dccda3d.json +96 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/yujiepan/qwen2.5-128k-tiny-random/4be7e684c15f704cb67d.json +78 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/yujiepan/qwen2.5-128k-tiny-random/6cead920a0dbb3daefb9.json +78 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/yujiepan/qwen2.5-128k-tiny-random/cd761b198c1bec21bd55.json +78 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3/Qwen/Qwen3-0.6B/2ed7a8812dbe9c7ab058.json +101 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/8fff447965fd3c157c0f.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/90f9e02faff8566070cb.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/f6c02365716b13dea692.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/smollm3/HuggingFaceTB/SmolLM3-3B/60de13af0adf5a679b2c.json +148 -0
- neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev3/qwen3_moe/Qwen/Qwen3-30B-A3B-Instruct-2507/fc72a21b6de27e9bcefe.json +79 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/compile_flags.json +1 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/model.done +0 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/model.hlo_module.pb +3 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/model.neff +0 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/compile_flags.json +1 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/model.done +0 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/model.hlo_module.pb +3 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/model.neff +3 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/wrapped_neff.hlo +3 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_0b67a734fc55d67768bd+253d6470/compile_flags.json +1 -0
- neuronxcc-2.19.8089.0+8ab9f450/MODULE_0b67a734fc55d67768bd+253d6470/model.done +0 -0
.gitattributes
CHANGED
@@ -10618,3 +10618,24 @@ neuronxcc-2.19.8089.0+8ab9f450/MODULE_0bac481d29ac1829a8af+253d6470/model.neff f
|
|
10618 |
neuronxcc-2.19.8089.0+8ab9f450/MODULE_2052cf656c210489741d+431f5505/model.neff filter=lfs diff=lfs merge=lfs -text
|
10619 |
neuronxcc-2.19.8089.0+8ab9f450/MODULE_b9a3b941bc75ca360b60+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10620 |
neuronxcc-2.19.8089.0+8ab9f450/MODULE_b9a3b941bc75ca360b60+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10618 |
neuronxcc-2.19.8089.0+8ab9f450/MODULE_2052cf656c210489741d+431f5505/model.neff filter=lfs diff=lfs merge=lfs -text
|
10619 |
neuronxcc-2.19.8089.0+8ab9f450/MODULE_b9a3b941bc75ca360b60+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10620 |
neuronxcc-2.19.8089.0+8ab9f450/MODULE_b9a3b941bc75ca360b60+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
10621 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10622 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
10623 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0b67a734fc55d67768bd+253d6470/model.neff filter=lfs diff=lfs merge=lfs -text
|
10624 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_184fcc045921c74845be+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10625 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_184fcc045921c74845be+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
10626 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_1d0c9d2357e7888e46cd+253d6470/model.neff filter=lfs diff=lfs merge=lfs -text
|
10627 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_20ce63f489314924a057+253d6470/model.neff filter=lfs diff=lfs merge=lfs -text
|
10628 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_40f6b4304be01d72d956+253d6470/model.neff filter=lfs diff=lfs merge=lfs -text
|
10629 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_64f5633fb25cfe2548d6+253d6470/model.neff filter=lfs diff=lfs merge=lfs -text
|
10630 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_806ef7cec3a07d99b574+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10631 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_806ef7cec3a07d99b574+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
10632 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_9f17b2a2f6b76b4ae4a8+253d6470/model.neff filter=lfs diff=lfs merge=lfs -text
|
10633 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_b522f7aabb6b091bde5f+253d6470/model.neff filter=lfs diff=lfs merge=lfs -text
|
10634 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_b72befbd04b196887e2e+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10635 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_b72befbd04b196887e2e+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
10636 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_c5daaeaa40996f359a5c+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10637 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_c5daaeaa40996f359a5c+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
10638 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_df21c2b35cb01f251b78+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10639 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_df21c2b35cb01f251b78+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
10640 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_f561bf7d12efdc67186d+cd3419b6/model.neff filter=lfs diff=lfs merge=lfs -text
|
10641 |
+
neuronxcc-2.19.8089.0+8ab9f450/MODULE_f561bf7d12efdc67186d+cd3419b6/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/Qwen/Qwen3-30B-A3B-Instruct-2507/2e075469bb7e246e19de.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 1,
|
11 |
+
"head_dim": 128,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 2048,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 6144,
|
16 |
+
"max_position_embeddings": 262144,
|
17 |
+
"max_window_layers": 48,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 768,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 8,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
|
29 |
+
"checkpoint_revision": "61082d4deaa4785f64943b443cbc2b5de7524fad",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 8,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 8,
|
40 |
+
"max_context_length": 4096,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 4096,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 4096,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 8,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 32,
|
65 |
+
"num_experts": 128,
|
66 |
+
"num_experts_per_tok": 8,
|
67 |
+
"num_hidden_layers": 48,
|
68 |
+
"num_key_value_heads": 4,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 10000000,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": false,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/63936687f0615be0c4b5.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
29 |
+
"checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 1024,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 1024,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 1024,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/67d11d4985c28ac7d5ff.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
29 |
+
"checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/cd03fa772b07fbb454c0.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
29 |
+
"checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/e5eddc28add59e2ee9d6.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 2,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
29 |
+
"checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 2,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/tiny-random/qwen3-moe/21c419fc8bcc3104cc37.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "tiny-random/qwen3-moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 2,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "tiny-random/qwen3-moe",
|
29 |
+
"checkpoint_revision": "10a349dcb488b10c27aa4a3c1dbefb74c41565c3",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 2,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/tiny-random/qwen3-moe/e14cd546176a864ee18b.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "tiny-random/qwen3-moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "tiny-random/qwen3-moe",
|
29 |
+
"checkpoint_revision": "10a349dcb488b10c27aa4a3c1dbefb74c41565c3",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/tiny-random/qwen3-moe/e99ce6fcf5be305e9424.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "tiny-random/qwen3-moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "tiny-random/qwen3-moe",
|
29 |
+
"checkpoint_revision": "10a349dcb488b10c27aa4a3c1dbefb74c41565c3",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/yujiepan/qwen3-moe-tiny-random/1cacc2cc42865a8b000d.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/qwen3-moe-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 2,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "yujiepan/qwen3-moe-tiny-random",
|
29 |
+
"checkpoint_revision": "fb6c5ee2a2c19bd9aced6d9afd8a858966a7bb7e",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 2,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/yujiepan/qwen3-moe-tiny-random/619f2f5356bdf0f59205.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/qwen3-moe-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "yujiepan/qwen3-moe-tiny-random",
|
29 |
+
"checkpoint_revision": "fb6c5ee2a2c19bd9aced6d9afd8a858966a7bb7e",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev1/qwen3_moe/yujiepan/qwen3-moe-tiny-random/67bc8c12b9b1221b38ad.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/qwen3-moe-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "yujiepan/qwen3-moe-tiny-random",
|
29 |
+
"checkpoint_revision": "fb6c5ee2a2c19bd9aced6d9afd8a858966a7bb7e",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev1",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/0f44022ffda5d90427b3.json
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"GraniteForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"attention_multiplier": 1.0,
|
11 |
+
"embedding_multiplier": 1.0,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 32,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 64,
|
16 |
+
"logits_scaling": 1.0,
|
17 |
+
"max_position_embeddings": 2048,
|
18 |
+
"mlp_bias": false,
|
19 |
+
"model_type": "granite",
|
20 |
+
"neuron": {
|
21 |
+
"_serialized_key": "NxDNeuronConfig",
|
22 |
+
"async_mode": false,
|
23 |
+
"attn_kernel_enabled": false,
|
24 |
+
"batch_size": 1,
|
25 |
+
"capacity_factor": null,
|
26 |
+
"cc_pipeline_tiling_factor": 2,
|
27 |
+
"checkpoint_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
|
28 |
+
"checkpoint_revision": "c3074ebc0ac2fe545305f5e5f6cce2cc9b2aa0c5",
|
29 |
+
"continuous_batching": false,
|
30 |
+
"enable_bucketing": false,
|
31 |
+
"ep_degree": 1,
|
32 |
+
"flash_decoding_enabled": false,
|
33 |
+
"fused_qkv": true,
|
34 |
+
"glu_mlp": true,
|
35 |
+
"is_chunked_prefill": false,
|
36 |
+
"local_ranks_size": 2,
|
37 |
+
"logical_nc_config": 1,
|
38 |
+
"max_batch_size": 1,
|
39 |
+
"max_context_length": 100,
|
40 |
+
"max_topk": 256,
|
41 |
+
"mlp_kernel_enabled": false,
|
42 |
+
"mlp_kernel_fuse_residual_add": false,
|
43 |
+
"n_active_tokens": 100,
|
44 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
45 |
+
"num_cores_per_group": 1,
|
46 |
+
"on_device_sampling": true,
|
47 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
48 |
+
"output_logits": false,
|
49 |
+
"padding_side": "right",
|
50 |
+
"pp_degree": 1,
|
51 |
+
"qkv_kernel_enabled": false,
|
52 |
+
"rpl_reduce_dtype": "bfloat16",
|
53 |
+
"sequence_length": 100,
|
54 |
+
"sequence_parallel_enabled": false,
|
55 |
+
"speculation_length": 0,
|
56 |
+
"start_rank_id": 0,
|
57 |
+
"target": null,
|
58 |
+
"torch_dtype": "bfloat16",
|
59 |
+
"tp_degree": 2,
|
60 |
+
"vocab_parallel": false
|
61 |
+
},
|
62 |
+
"num_attention_heads": 4,
|
63 |
+
"num_hidden_layers": 2,
|
64 |
+
"num_key_value_heads": 4,
|
65 |
+
"residual_multiplier": 1.0,
|
66 |
+
"rms_norm_eps": 1e-06,
|
67 |
+
"rope_scaling": null,
|
68 |
+
"rope_theta": 10000.0,
|
69 |
+
"tie_word_embeddings": false,
|
70 |
+
"use_cache": true,
|
71 |
+
"vocab_size": 49152
|
72 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/1471022e22b5b8b3de3b.json
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"GraniteForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"attention_multiplier": 1.0,
|
11 |
+
"embedding_multiplier": 1.0,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 32,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 64,
|
16 |
+
"logits_scaling": 1.0,
|
17 |
+
"max_position_embeddings": 2048,
|
18 |
+
"mlp_bias": false,
|
19 |
+
"model_type": "granite",
|
20 |
+
"neuron": {
|
21 |
+
"_serialized_key": "NxDNeuronConfig",
|
22 |
+
"async_mode": false,
|
23 |
+
"attn_kernel_enabled": false,
|
24 |
+
"batch_size": 1,
|
25 |
+
"capacity_factor": null,
|
26 |
+
"cc_pipeline_tiling_factor": 2,
|
27 |
+
"checkpoint_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
|
28 |
+
"checkpoint_revision": "c3074ebc0ac2fe545305f5e5f6cce2cc9b2aa0c5",
|
29 |
+
"continuous_batching": false,
|
30 |
+
"enable_bucketing": false,
|
31 |
+
"ep_degree": 1,
|
32 |
+
"flash_decoding_enabled": false,
|
33 |
+
"fused_qkv": true,
|
34 |
+
"glu_mlp": true,
|
35 |
+
"is_chunked_prefill": false,
|
36 |
+
"local_ranks_size": 2,
|
37 |
+
"logical_nc_config": 1,
|
38 |
+
"max_batch_size": 1,
|
39 |
+
"max_context_length": 100,
|
40 |
+
"max_topk": 256,
|
41 |
+
"mlp_kernel_enabled": false,
|
42 |
+
"mlp_kernel_fuse_residual_add": false,
|
43 |
+
"n_active_tokens": 100,
|
44 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
45 |
+
"num_cores_per_group": 1,
|
46 |
+
"on_device_sampling": true,
|
47 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
48 |
+
"output_logits": false,
|
49 |
+
"padding_side": "right",
|
50 |
+
"pp_degree": 1,
|
51 |
+
"qkv_kernel_enabled": false,
|
52 |
+
"rpl_reduce_dtype": "float16",
|
53 |
+
"sequence_length": 100,
|
54 |
+
"sequence_parallel_enabled": false,
|
55 |
+
"speculation_length": 0,
|
56 |
+
"start_rank_id": 0,
|
57 |
+
"target": null,
|
58 |
+
"torch_dtype": "float16",
|
59 |
+
"tp_degree": 2,
|
60 |
+
"vocab_parallel": false
|
61 |
+
},
|
62 |
+
"num_attention_heads": 4,
|
63 |
+
"num_hidden_layers": 2,
|
64 |
+
"num_key_value_heads": 4,
|
65 |
+
"residual_multiplier": 1.0,
|
66 |
+
"rms_norm_eps": 1e-06,
|
67 |
+
"rope_scaling": null,
|
68 |
+
"rope_theta": 10000.0,
|
69 |
+
"tie_word_embeddings": false,
|
70 |
+
"use_cache": true,
|
71 |
+
"vocab_size": 49152
|
72 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/c9c911d79dbddb873dcf.json
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"GraniteForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"attention_multiplier": 1.0,
|
11 |
+
"embedding_multiplier": 1.0,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 32,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 64,
|
16 |
+
"logits_scaling": 1.0,
|
17 |
+
"max_position_embeddings": 2048,
|
18 |
+
"mlp_bias": false,
|
19 |
+
"model_type": "granite",
|
20 |
+
"neuron": {
|
21 |
+
"_serialized_key": "NxDNeuronConfig",
|
22 |
+
"async_mode": false,
|
23 |
+
"attn_kernel_enabled": false,
|
24 |
+
"batch_size": 2,
|
25 |
+
"capacity_factor": null,
|
26 |
+
"cc_pipeline_tiling_factor": 2,
|
27 |
+
"checkpoint_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
|
28 |
+
"checkpoint_revision": "c3074ebc0ac2fe545305f5e5f6cce2cc9b2aa0c5",
|
29 |
+
"continuous_batching": true,
|
30 |
+
"enable_bucketing": false,
|
31 |
+
"ep_degree": 1,
|
32 |
+
"flash_decoding_enabled": false,
|
33 |
+
"fused_qkv": true,
|
34 |
+
"glu_mlp": true,
|
35 |
+
"is_chunked_prefill": false,
|
36 |
+
"local_ranks_size": 2,
|
37 |
+
"logical_nc_config": 1,
|
38 |
+
"max_batch_size": 2,
|
39 |
+
"max_context_length": 100,
|
40 |
+
"max_topk": 256,
|
41 |
+
"mlp_kernel_enabled": false,
|
42 |
+
"mlp_kernel_fuse_residual_add": false,
|
43 |
+
"n_active_tokens": 100,
|
44 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
45 |
+
"num_cores_per_group": 1,
|
46 |
+
"on_device_sampling": true,
|
47 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
48 |
+
"output_logits": false,
|
49 |
+
"padding_side": "right",
|
50 |
+
"pp_degree": 1,
|
51 |
+
"qkv_kernel_enabled": false,
|
52 |
+
"rpl_reduce_dtype": "float16",
|
53 |
+
"sequence_length": 100,
|
54 |
+
"sequence_parallel_enabled": false,
|
55 |
+
"speculation_length": 0,
|
56 |
+
"start_rank_id": 0,
|
57 |
+
"target": null,
|
58 |
+
"torch_dtype": "float16",
|
59 |
+
"tp_degree": 2,
|
60 |
+
"vocab_parallel": false
|
61 |
+
},
|
62 |
+
"num_attention_heads": 4,
|
63 |
+
"num_hidden_layers": 2,
|
64 |
+
"num_key_value_heads": 4,
|
65 |
+
"residual_multiplier": 1.0,
|
66 |
+
"rms_norm_eps": 1e-06,
|
67 |
+
"rope_scaling": null,
|
68 |
+
"rope_theta": 10000.0,
|
69 |
+
"tie_word_embeddings": false,
|
70 |
+
"use_cache": true,
|
71 |
+
"vocab_size": 49152
|
72 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/granite/ibm-granite/granite-3.1-2b-instruct/b66dbd045f3e73eb7427.json
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "ibm-granite/granite-3.1-2b-instruct",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"GraniteForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.1,
|
10 |
+
"attention_multiplier": 0.015625,
|
11 |
+
"embedding_multiplier": 12.0,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 2048,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 8192,
|
16 |
+
"logits_scaling": 8.0,
|
17 |
+
"max_position_embeddings": 131072,
|
18 |
+
"mlp_bias": false,
|
19 |
+
"model_type": "granite",
|
20 |
+
"neuron": {
|
21 |
+
"_serialized_key": "NxDNeuronConfig",
|
22 |
+
"async_mode": false,
|
23 |
+
"attn_kernel_enabled": false,
|
24 |
+
"batch_size": 4,
|
25 |
+
"capacity_factor": null,
|
26 |
+
"cc_pipeline_tiling_factor": 2,
|
27 |
+
"checkpoint_id": "ibm-granite/granite-3.1-2b-instruct",
|
28 |
+
"checkpoint_revision": "bbc2aed595bd38bd770263dc3ab831db9794441d",
|
29 |
+
"continuous_batching": true,
|
30 |
+
"enable_bucketing": false,
|
31 |
+
"ep_degree": 1,
|
32 |
+
"flash_decoding_enabled": false,
|
33 |
+
"fused_qkv": true,
|
34 |
+
"glu_mlp": true,
|
35 |
+
"is_chunked_prefill": false,
|
36 |
+
"local_ranks_size": 2,
|
37 |
+
"logical_nc_config": 1,
|
38 |
+
"max_batch_size": 4,
|
39 |
+
"max_context_length": 4096,
|
40 |
+
"max_topk": 256,
|
41 |
+
"mlp_kernel_enabled": false,
|
42 |
+
"mlp_kernel_fuse_residual_add": false,
|
43 |
+
"n_active_tokens": 4096,
|
44 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
45 |
+
"num_cores_per_group": 1,
|
46 |
+
"on_device_sampling": true,
|
47 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
48 |
+
"output_logits": false,
|
49 |
+
"padding_side": "right",
|
50 |
+
"pp_degree": 1,
|
51 |
+
"qkv_kernel_enabled": false,
|
52 |
+
"rpl_reduce_dtype": "bfloat16",
|
53 |
+
"sequence_length": 4096,
|
54 |
+
"sequence_parallel_enabled": false,
|
55 |
+
"speculation_length": 0,
|
56 |
+
"start_rank_id": 0,
|
57 |
+
"target": null,
|
58 |
+
"torch_dtype": "bfloat16",
|
59 |
+
"tp_degree": 2,
|
60 |
+
"vocab_parallel": false
|
61 |
+
},
|
62 |
+
"num_attention_heads": 32,
|
63 |
+
"num_hidden_layers": 40,
|
64 |
+
"num_key_value_heads": 8,
|
65 |
+
"residual_multiplier": 0.22,
|
66 |
+
"rms_norm_eps": 1e-05,
|
67 |
+
"rope_scaling": null,
|
68 |
+
"rope_theta": 5000000.0,
|
69 |
+
"tie_word_embeddings": true,
|
70 |
+
"use_cache": true,
|
71 |
+
"vocab_size": 49155
|
72 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/20b7299fc104cc69ce0f.json
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "llamafactory/tiny-random-Llama-3",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"LlamaForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"head_dim": 4,
|
11 |
+
"hidden_act": "silu",
|
12 |
+
"hidden_size": 16,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 64,
|
15 |
+
"max_position_embeddings": 131072,
|
16 |
+
"mlp_bias": false,
|
17 |
+
"model_type": "llama",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 1,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "llamafactory/tiny-random-Llama-3",
|
26 |
+
"checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
|
27 |
+
"continuous_batching": false,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 1,
|
37 |
+
"max_context_length": 100,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 100,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "bfloat16",
|
51 |
+
"sequence_length": 100,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "bfloat16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 4,
|
61 |
+
"num_hidden_layers": 2,
|
62 |
+
"num_key_value_heads": 4,
|
63 |
+
"pretraining_tp": 1,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_scaling": {
|
66 |
+
"factor": 8.0,
|
67 |
+
"high_freq_factor": 4.0,
|
68 |
+
"low_freq_factor": 1.0,
|
69 |
+
"original_max_position_embeddings": 8192,
|
70 |
+
"rope_type": "llama3"
|
71 |
+
},
|
72 |
+
"rope_theta": 500000.0,
|
73 |
+
"tie_word_embeddings": false,
|
74 |
+
"use_cache": true,
|
75 |
+
"vocab_size": 128256
|
76 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/502c5523ff7585cd3287.json
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "llamafactory/tiny-random-Llama-3",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"LlamaForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"head_dim": 4,
|
11 |
+
"hidden_act": "silu",
|
12 |
+
"hidden_size": 16,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 64,
|
15 |
+
"max_position_embeddings": 131072,
|
16 |
+
"mlp_bias": false,
|
17 |
+
"model_type": "llama",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 1,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "llamafactory/tiny-random-Llama-3",
|
26 |
+
"checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
|
27 |
+
"continuous_batching": false,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 1,
|
37 |
+
"max_context_length": 131072,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 131072,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "bfloat16",
|
51 |
+
"sequence_length": 131072,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "bfloat16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 4,
|
61 |
+
"num_hidden_layers": 2,
|
62 |
+
"num_key_value_heads": 4,
|
63 |
+
"pretraining_tp": 1,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_scaling": {
|
66 |
+
"factor": 8.0,
|
67 |
+
"high_freq_factor": 4.0,
|
68 |
+
"low_freq_factor": 1.0,
|
69 |
+
"original_max_position_embeddings": 8192,
|
70 |
+
"rope_type": "llama3"
|
71 |
+
},
|
72 |
+
"rope_theta": 500000.0,
|
73 |
+
"tie_word_embeddings": false,
|
74 |
+
"use_cache": true,
|
75 |
+
"vocab_size": 128256
|
76 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/9c27c3f4a94313192c63.json
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "llamafactory/tiny-random-Llama-3",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"LlamaForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"head_dim": 4,
|
11 |
+
"hidden_act": "silu",
|
12 |
+
"hidden_size": 16,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 64,
|
15 |
+
"max_position_embeddings": 131072,
|
16 |
+
"mlp_bias": false,
|
17 |
+
"model_type": "llama",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 1,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "llamafactory/tiny-random-Llama-3",
|
26 |
+
"checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
|
27 |
+
"continuous_batching": false,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 1,
|
37 |
+
"max_context_length": 100,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 100,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "float16",
|
51 |
+
"sequence_length": 100,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "float16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 4,
|
61 |
+
"num_hidden_layers": 2,
|
62 |
+
"num_key_value_heads": 4,
|
63 |
+
"pretraining_tp": 1,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_scaling": {
|
66 |
+
"factor": 8.0,
|
67 |
+
"high_freq_factor": 4.0,
|
68 |
+
"low_freq_factor": 1.0,
|
69 |
+
"original_max_position_embeddings": 8192,
|
70 |
+
"rope_type": "llama3"
|
71 |
+
},
|
72 |
+
"rope_theta": 500000.0,
|
73 |
+
"tie_word_embeddings": false,
|
74 |
+
"use_cache": true,
|
75 |
+
"vocab_size": 128256
|
76 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/llamafactory/tiny-random-Llama-3/f0cd6ae3a57125c25463.json
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "llamafactory/tiny-random-Llama-3",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"LlamaForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"head_dim": 4,
|
11 |
+
"hidden_act": "silu",
|
12 |
+
"hidden_size": 16,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 64,
|
15 |
+
"max_position_embeddings": 131072,
|
16 |
+
"mlp_bias": false,
|
17 |
+
"model_type": "llama",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 2,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "llamafactory/tiny-random-Llama-3",
|
26 |
+
"checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
|
27 |
+
"continuous_batching": true,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 2,
|
37 |
+
"max_context_length": 100,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 100,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "float16",
|
51 |
+
"sequence_length": 100,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "float16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 4,
|
61 |
+
"num_hidden_layers": 2,
|
62 |
+
"num_key_value_heads": 4,
|
63 |
+
"pretraining_tp": 1,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_scaling": {
|
66 |
+
"factor": 8.0,
|
67 |
+
"high_freq_factor": 4.0,
|
68 |
+
"low_freq_factor": 1.0,
|
69 |
+
"original_max_position_embeddings": 8192,
|
70 |
+
"rope_type": "llama3"
|
71 |
+
},
|
72 |
+
"rope_theta": 500000.0,
|
73 |
+
"tie_word_embeddings": false,
|
74 |
+
"use_cache": true,
|
75 |
+
"vocab_size": 128256
|
76 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/llama/unsloth/Llama-3.2-1B-Instruct/d4cdf18983e9784091ec.json
ADDED
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "unsloth/Llama-3.2-1B-Instruct",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"LlamaForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"head_dim": 64,
|
11 |
+
"hidden_act": "silu",
|
12 |
+
"hidden_size": 2048,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 8192,
|
15 |
+
"max_position_embeddings": 131072,
|
16 |
+
"mlp_bias": false,
|
17 |
+
"model_type": "llama",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 4,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "unsloth/Llama-3.2-1B-Instruct",
|
26 |
+
"checkpoint_revision": "5a8abab4a5d6f164389b1079fb721cfab8d7126c",
|
27 |
+
"continuous_batching": true,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 4,
|
37 |
+
"max_context_length": 4096,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 4096,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "float16",
|
51 |
+
"sequence_length": 4096,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "float16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 32,
|
61 |
+
"num_hidden_layers": 16,
|
62 |
+
"num_key_value_heads": 8,
|
63 |
+
"pretraining_tp": 1,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_scaling": {
|
66 |
+
"factor": 32.0,
|
67 |
+
"high_freq_factor": 4.0,
|
68 |
+
"low_freq_factor": 1.0,
|
69 |
+
"original_max_position_embeddings": 8192,
|
70 |
+
"rope_type": "llama3"
|
71 |
+
},
|
72 |
+
"rope_theta": 500000.0,
|
73 |
+
"tie_word_embeddings": true,
|
74 |
+
"unsloth_fixed": true,
|
75 |
+
"use_cache": true,
|
76 |
+
"vocab_size": 128256
|
77 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/mixtral/dacorvo/Mixtral-tiny/34134d6beddf69d6e6c0.json
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "dacorvo/Mixtral-tiny",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"MixtralForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"head_dim": 32,
|
10 |
+
"hidden_act": "silu",
|
11 |
+
"hidden_size": 1024,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 3584,
|
14 |
+
"max_position_embeddings": 1024,
|
15 |
+
"model_type": "mixtral",
|
16 |
+
"neuron": {
|
17 |
+
"_serialized_key": "NxDNeuronConfig",
|
18 |
+
"async_mode": false,
|
19 |
+
"attn_kernel_enabled": false,
|
20 |
+
"batch_size": 1,
|
21 |
+
"capacity_factor": null,
|
22 |
+
"cc_pipeline_tiling_factor": 2,
|
23 |
+
"checkpoint_id": "dacorvo/Mixtral-tiny",
|
24 |
+
"checkpoint_revision": "c557ba205ddff6ea911f4719e0d543d6c08356b6",
|
25 |
+
"continuous_batching": false,
|
26 |
+
"enable_bucketing": false,
|
27 |
+
"ep_degree": 1,
|
28 |
+
"flash_decoding_enabled": false,
|
29 |
+
"fused_qkv": false,
|
30 |
+
"glu_mlp": true,
|
31 |
+
"is_chunked_prefill": false,
|
32 |
+
"local_ranks_size": 2,
|
33 |
+
"logical_nc_config": 1,
|
34 |
+
"max_batch_size": 1,
|
35 |
+
"max_context_length": 100,
|
36 |
+
"max_topk": 256,
|
37 |
+
"mlp_kernel_enabled": false,
|
38 |
+
"mlp_kernel_fuse_residual_add": false,
|
39 |
+
"n_active_tokens": 100,
|
40 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
41 |
+
"num_cores_per_group": 1,
|
42 |
+
"on_device_sampling": false,
|
43 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
44 |
+
"output_logits": false,
|
45 |
+
"padding_side": "right",
|
46 |
+
"pp_degree": 1,
|
47 |
+
"qkv_kernel_enabled": false,
|
48 |
+
"rpl_reduce_dtype": "float16",
|
49 |
+
"sequence_length": 100,
|
50 |
+
"sequence_parallel_enabled": false,
|
51 |
+
"speculation_length": 0,
|
52 |
+
"start_rank_id": 0,
|
53 |
+
"target": null,
|
54 |
+
"torch_dtype": "float16",
|
55 |
+
"tp_degree": 2,
|
56 |
+
"vocab_parallel": false
|
57 |
+
},
|
58 |
+
"num_attention_heads": 32,
|
59 |
+
"num_experts_per_tok": 2,
|
60 |
+
"num_hidden_layers": 2,
|
61 |
+
"num_key_value_heads": 8,
|
62 |
+
"num_local_experts": 8,
|
63 |
+
"output_router_logits": false,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_theta": 10000.0,
|
66 |
+
"router_aux_loss_coef": 0.001,
|
67 |
+
"router_jitter_noise": 0.0,
|
68 |
+
"sliding_window": 4096,
|
69 |
+
"tie_word_embeddings": false,
|
70 |
+
"use_cache": true,
|
71 |
+
"vocab_size": 32000
|
72 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/mixtral/dacorvo/Mixtral-tiny/4625dfce3a6ec89cfa42.json
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "dacorvo/Mixtral-tiny",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"MixtralForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"head_dim": 32,
|
10 |
+
"hidden_act": "silu",
|
11 |
+
"hidden_size": 1024,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 3584,
|
14 |
+
"max_position_embeddings": 1024,
|
15 |
+
"model_type": "mixtral",
|
16 |
+
"neuron": {
|
17 |
+
"_serialized_key": "NxDNeuronConfig",
|
18 |
+
"async_mode": false,
|
19 |
+
"attn_kernel_enabled": false,
|
20 |
+
"batch_size": 1,
|
21 |
+
"capacity_factor": null,
|
22 |
+
"cc_pipeline_tiling_factor": 2,
|
23 |
+
"checkpoint_id": "dacorvo/Mixtral-tiny",
|
24 |
+
"checkpoint_revision": "c557ba205ddff6ea911f4719e0d543d6c08356b6",
|
25 |
+
"continuous_batching": false,
|
26 |
+
"enable_bucketing": false,
|
27 |
+
"ep_degree": 1,
|
28 |
+
"flash_decoding_enabled": false,
|
29 |
+
"fused_qkv": false,
|
30 |
+
"glu_mlp": true,
|
31 |
+
"is_chunked_prefill": false,
|
32 |
+
"local_ranks_size": 2,
|
33 |
+
"logical_nc_config": 1,
|
34 |
+
"max_batch_size": 1,
|
35 |
+
"max_context_length": 100,
|
36 |
+
"max_topk": 256,
|
37 |
+
"mlp_kernel_enabled": false,
|
38 |
+
"mlp_kernel_fuse_residual_add": false,
|
39 |
+
"n_active_tokens": 100,
|
40 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
41 |
+
"num_cores_per_group": 1,
|
42 |
+
"on_device_sampling": false,
|
43 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
44 |
+
"output_logits": false,
|
45 |
+
"padding_side": "right",
|
46 |
+
"pp_degree": 1,
|
47 |
+
"qkv_kernel_enabled": false,
|
48 |
+
"rpl_reduce_dtype": "bfloat16",
|
49 |
+
"sequence_length": 100,
|
50 |
+
"sequence_parallel_enabled": false,
|
51 |
+
"speculation_length": 0,
|
52 |
+
"start_rank_id": 0,
|
53 |
+
"target": null,
|
54 |
+
"torch_dtype": "bfloat16",
|
55 |
+
"tp_degree": 2,
|
56 |
+
"vocab_parallel": false
|
57 |
+
},
|
58 |
+
"num_attention_heads": 32,
|
59 |
+
"num_experts_per_tok": 2,
|
60 |
+
"num_hidden_layers": 2,
|
61 |
+
"num_key_value_heads": 8,
|
62 |
+
"num_local_experts": 8,
|
63 |
+
"output_router_logits": false,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_theta": 10000.0,
|
66 |
+
"router_aux_loss_coef": 0.001,
|
67 |
+
"router_jitter_noise": 0.0,
|
68 |
+
"sliding_window": 4096,
|
69 |
+
"tie_word_embeddings": false,
|
70 |
+
"use_cache": true,
|
71 |
+
"vocab_size": 32000
|
72 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/mixtral/dacorvo/Mixtral-tiny/f790967cf96542a801bc.json
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "dacorvo/Mixtral-tiny",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"MixtralForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"head_dim": 32,
|
10 |
+
"hidden_act": "silu",
|
11 |
+
"hidden_size": 1024,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 3584,
|
14 |
+
"max_position_embeddings": 1024,
|
15 |
+
"model_type": "mixtral",
|
16 |
+
"neuron": {
|
17 |
+
"_serialized_key": "NxDNeuronConfig",
|
18 |
+
"async_mode": false,
|
19 |
+
"attn_kernel_enabled": false,
|
20 |
+
"batch_size": 2,
|
21 |
+
"capacity_factor": null,
|
22 |
+
"cc_pipeline_tiling_factor": 2,
|
23 |
+
"checkpoint_id": "dacorvo/Mixtral-tiny",
|
24 |
+
"checkpoint_revision": "c557ba205ddff6ea911f4719e0d543d6c08356b6",
|
25 |
+
"continuous_batching": false,
|
26 |
+
"enable_bucketing": false,
|
27 |
+
"ep_degree": 1,
|
28 |
+
"flash_decoding_enabled": false,
|
29 |
+
"fused_qkv": false,
|
30 |
+
"glu_mlp": true,
|
31 |
+
"is_chunked_prefill": false,
|
32 |
+
"local_ranks_size": 2,
|
33 |
+
"logical_nc_config": 1,
|
34 |
+
"max_batch_size": 2,
|
35 |
+
"max_context_length": 100,
|
36 |
+
"max_topk": 256,
|
37 |
+
"mlp_kernel_enabled": false,
|
38 |
+
"mlp_kernel_fuse_residual_add": false,
|
39 |
+
"n_active_tokens": 100,
|
40 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
41 |
+
"num_cores_per_group": 1,
|
42 |
+
"on_device_sampling": false,
|
43 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
44 |
+
"output_logits": false,
|
45 |
+
"padding_side": "right",
|
46 |
+
"pp_degree": 1,
|
47 |
+
"qkv_kernel_enabled": false,
|
48 |
+
"rpl_reduce_dtype": "float16",
|
49 |
+
"sequence_length": 100,
|
50 |
+
"sequence_parallel_enabled": false,
|
51 |
+
"speculation_length": 0,
|
52 |
+
"start_rank_id": 0,
|
53 |
+
"target": null,
|
54 |
+
"torch_dtype": "float16",
|
55 |
+
"tp_degree": 2,
|
56 |
+
"vocab_parallel": false
|
57 |
+
},
|
58 |
+
"num_attention_heads": 32,
|
59 |
+
"num_experts_per_tok": 2,
|
60 |
+
"num_hidden_layers": 2,
|
61 |
+
"num_key_value_heads": 8,
|
62 |
+
"num_local_experts": 8,
|
63 |
+
"output_router_logits": false,
|
64 |
+
"rms_norm_eps": 1e-05,
|
65 |
+
"rope_theta": 10000.0,
|
66 |
+
"router_aux_loss_coef": 0.001,
|
67 |
+
"router_jitter_noise": 0.0,
|
68 |
+
"sliding_window": 4096,
|
69 |
+
"tie_word_embeddings": false,
|
70 |
+
"use_cache": true,
|
71 |
+
"vocab_size": 32000
|
72 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/microsoft/Phi-3-mini-4k-instruct/c3b4f930a6c51ad40028.json
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "microsoft/Phi-3-mini-4k-instruct",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Phi3ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"auto_map": {
|
11 |
+
"AutoConfig": "configuration_phi3.Phi3Config",
|
12 |
+
"AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
|
13 |
+
},
|
14 |
+
"embd_pdrop": 0.0,
|
15 |
+
"hidden_act": "silu",
|
16 |
+
"hidden_size": 3072,
|
17 |
+
"initializer_range": 0.02,
|
18 |
+
"intermediate_size": 8192,
|
19 |
+
"max_position_embeddings": 4096,
|
20 |
+
"model_type": "phi3",
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 4,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "microsoft/Phi-3-mini-4k-instruct",
|
29 |
+
"checkpoint_revision": "0a67737cc96d2554230f90338b163bc6380a2a85",
|
30 |
+
"continuous_batching": true,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": true,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 4,
|
40 |
+
"max_context_length": 4096,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 4096,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": true,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 4096,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"num_attention_heads": 32,
|
64 |
+
"num_hidden_layers": 32,
|
65 |
+
"num_key_value_heads": 32,
|
66 |
+
"original_max_position_embeddings": 4096,
|
67 |
+
"partial_rotary_factor": 1.0,
|
68 |
+
"resid_pdrop": 0.0,
|
69 |
+
"rms_norm_eps": 1e-05,
|
70 |
+
"rope_scaling": null,
|
71 |
+
"rope_theta": 10000.0,
|
72 |
+
"sliding_window": 2047,
|
73 |
+
"tie_word_embeddings": false,
|
74 |
+
"use_cache": true,
|
75 |
+
"vocab_size": 32064
|
76 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/yujiepan/phi-4-tiny-random/63d7815d3de40f55a3ae.json
ADDED
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/phi-4-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Phi3ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"auto_map": {},
|
11 |
+
"embd_pdrop": 0.0,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 16,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 32,
|
16 |
+
"max_position_embeddings": 16384,
|
17 |
+
"model_type": "phi3",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 2,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "yujiepan/phi-4-tiny-random",
|
26 |
+
"checkpoint_revision": "18a9a1168dc97ac6d128f811925670c275610f5a",
|
27 |
+
"continuous_batching": true,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 2,
|
37 |
+
"max_context_length": 100,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 100,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "float16",
|
51 |
+
"sequence_length": 100,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "float16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 2,
|
61 |
+
"num_hidden_layers": 2,
|
62 |
+
"num_key_value_heads": 1,
|
63 |
+
"original_max_position_embeddings": 16384,
|
64 |
+
"partial_rotary_factor": 1.0,
|
65 |
+
"resid_pdrop": 0.0,
|
66 |
+
"rms_norm_eps": 1e-05,
|
67 |
+
"rope_scaling": null,
|
68 |
+
"rope_theta": 250000,
|
69 |
+
"sliding_window": null,
|
70 |
+
"tie_word_embeddings": false,
|
71 |
+
"use_cache": true,
|
72 |
+
"vocab_size": 100352
|
73 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/yujiepan/phi-4-tiny-random/892d9f2862662e5407c6.json
ADDED
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/phi-4-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Phi3ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"auto_map": {},
|
11 |
+
"embd_pdrop": 0.0,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 16,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 32,
|
16 |
+
"max_position_embeddings": 16384,
|
17 |
+
"model_type": "phi3",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 1,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "yujiepan/phi-4-tiny-random",
|
26 |
+
"checkpoint_revision": "18a9a1168dc97ac6d128f811925670c275610f5a",
|
27 |
+
"continuous_batching": false,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 1,
|
37 |
+
"max_context_length": 100,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 100,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "float16",
|
51 |
+
"sequence_length": 100,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "float16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 2,
|
61 |
+
"num_hidden_layers": 2,
|
62 |
+
"num_key_value_heads": 1,
|
63 |
+
"original_max_position_embeddings": 16384,
|
64 |
+
"partial_rotary_factor": 1.0,
|
65 |
+
"resid_pdrop": 0.0,
|
66 |
+
"rms_norm_eps": 1e-05,
|
67 |
+
"rope_scaling": null,
|
68 |
+
"rope_theta": 250000,
|
69 |
+
"sliding_window": null,
|
70 |
+
"tie_word_embeddings": false,
|
71 |
+
"use_cache": true,
|
72 |
+
"vocab_size": 100352
|
73 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/phi3/yujiepan/phi-4-tiny-random/e97f2d9bfc450d08ef3b.json
ADDED
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/phi-4-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Phi3ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"auto_map": {},
|
11 |
+
"embd_pdrop": 0.0,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 16,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 32,
|
16 |
+
"max_position_embeddings": 16384,
|
17 |
+
"model_type": "phi3",
|
18 |
+
"neuron": {
|
19 |
+
"_serialized_key": "NxDNeuronConfig",
|
20 |
+
"async_mode": false,
|
21 |
+
"attn_kernel_enabled": false,
|
22 |
+
"batch_size": 1,
|
23 |
+
"capacity_factor": null,
|
24 |
+
"cc_pipeline_tiling_factor": 2,
|
25 |
+
"checkpoint_id": "yujiepan/phi-4-tiny-random",
|
26 |
+
"checkpoint_revision": "18a9a1168dc97ac6d128f811925670c275610f5a",
|
27 |
+
"continuous_batching": false,
|
28 |
+
"enable_bucketing": false,
|
29 |
+
"ep_degree": 1,
|
30 |
+
"flash_decoding_enabled": false,
|
31 |
+
"fused_qkv": true,
|
32 |
+
"glu_mlp": true,
|
33 |
+
"is_chunked_prefill": false,
|
34 |
+
"local_ranks_size": 2,
|
35 |
+
"logical_nc_config": 1,
|
36 |
+
"max_batch_size": 1,
|
37 |
+
"max_context_length": 100,
|
38 |
+
"max_topk": 256,
|
39 |
+
"mlp_kernel_enabled": false,
|
40 |
+
"mlp_kernel_fuse_residual_add": false,
|
41 |
+
"n_active_tokens": 100,
|
42 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
43 |
+
"num_cores_per_group": 1,
|
44 |
+
"on_device_sampling": true,
|
45 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
46 |
+
"output_logits": false,
|
47 |
+
"padding_side": "right",
|
48 |
+
"pp_degree": 1,
|
49 |
+
"qkv_kernel_enabled": false,
|
50 |
+
"rpl_reduce_dtype": "bfloat16",
|
51 |
+
"sequence_length": 100,
|
52 |
+
"sequence_parallel_enabled": false,
|
53 |
+
"speculation_length": 0,
|
54 |
+
"start_rank_id": 0,
|
55 |
+
"target": null,
|
56 |
+
"torch_dtype": "bfloat16",
|
57 |
+
"tp_degree": 2,
|
58 |
+
"vocab_parallel": false
|
59 |
+
},
|
60 |
+
"num_attention_heads": 2,
|
61 |
+
"num_hidden_layers": 2,
|
62 |
+
"num_key_value_heads": 1,
|
63 |
+
"original_max_position_embeddings": 16384,
|
64 |
+
"partial_rotary_factor": 1.0,
|
65 |
+
"resid_pdrop": 0.0,
|
66 |
+
"rms_norm_eps": 1e-05,
|
67 |
+
"rope_scaling": null,
|
68 |
+
"rope_theta": 250000,
|
69 |
+
"sliding_window": null,
|
70 |
+
"tie_word_embeddings": false,
|
71 |
+
"use_cache": true,
|
72 |
+
"vocab_size": 100352
|
73 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/Qwen/Qwen2.5-0.5B/00b909b9addfb3e82c75.json
ADDED
@@ -0,0 +1,96 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "Qwen/Qwen2.5-0.5B",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen2ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 896,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 4864,
|
13 |
+
"layer_types": [
|
14 |
+
"full_attention",
|
15 |
+
"full_attention",
|
16 |
+
"full_attention",
|
17 |
+
"full_attention",
|
18 |
+
"full_attention",
|
19 |
+
"full_attention",
|
20 |
+
"full_attention",
|
21 |
+
"full_attention",
|
22 |
+
"full_attention",
|
23 |
+
"full_attention",
|
24 |
+
"full_attention",
|
25 |
+
"full_attention",
|
26 |
+
"full_attention",
|
27 |
+
"full_attention",
|
28 |
+
"full_attention",
|
29 |
+
"full_attention",
|
30 |
+
"full_attention",
|
31 |
+
"full_attention",
|
32 |
+
"full_attention",
|
33 |
+
"full_attention",
|
34 |
+
"full_attention",
|
35 |
+
"full_attention",
|
36 |
+
"full_attention",
|
37 |
+
"full_attention"
|
38 |
+
],
|
39 |
+
"max_position_embeddings": 32768,
|
40 |
+
"max_window_layers": 24,
|
41 |
+
"model_type": "qwen2",
|
42 |
+
"neuron": {
|
43 |
+
"_serialized_key": "NxDNeuronConfig",
|
44 |
+
"async_mode": false,
|
45 |
+
"attn_kernel_enabled": false,
|
46 |
+
"batch_size": 4,
|
47 |
+
"capacity_factor": null,
|
48 |
+
"cc_pipeline_tiling_factor": 2,
|
49 |
+
"checkpoint_id": "Qwen/Qwen2.5-0.5B",
|
50 |
+
"checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
|
51 |
+
"continuous_batching": true,
|
52 |
+
"enable_bucketing": false,
|
53 |
+
"ep_degree": 1,
|
54 |
+
"flash_decoding_enabled": false,
|
55 |
+
"fused_qkv": false,
|
56 |
+
"glu_mlp": true,
|
57 |
+
"is_chunked_prefill": false,
|
58 |
+
"local_ranks_size": 2,
|
59 |
+
"logical_nc_config": 1,
|
60 |
+
"max_batch_size": 4,
|
61 |
+
"max_context_length": 4096,
|
62 |
+
"max_topk": 256,
|
63 |
+
"mlp_kernel_enabled": false,
|
64 |
+
"mlp_kernel_fuse_residual_add": false,
|
65 |
+
"n_active_tokens": 4096,
|
66 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
67 |
+
"num_cores_per_group": 1,
|
68 |
+
"on_device_sampling": false,
|
69 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
70 |
+
"output_logits": false,
|
71 |
+
"padding_side": "right",
|
72 |
+
"pp_degree": 1,
|
73 |
+
"qkv_kernel_enabled": false,
|
74 |
+
"rpl_reduce_dtype": "float16",
|
75 |
+
"sequence_length": 4096,
|
76 |
+
"sequence_parallel_enabled": false,
|
77 |
+
"speculation_length": 0,
|
78 |
+
"start_rank_id": 0,
|
79 |
+
"target": null,
|
80 |
+
"torch_dtype": "float16",
|
81 |
+
"tp_degree": 2,
|
82 |
+
"vocab_parallel": false
|
83 |
+
},
|
84 |
+
"num_attention_heads": 14,
|
85 |
+
"num_hidden_layers": 24,
|
86 |
+
"num_key_value_heads": 2,
|
87 |
+
"rms_norm_eps": 1e-06,
|
88 |
+
"rope_scaling": null,
|
89 |
+
"rope_theta": 1000000.0,
|
90 |
+
"sliding_window": null,
|
91 |
+
"tie_word_embeddings": true,
|
92 |
+
"use_cache": true,
|
93 |
+
"use_mrope": false,
|
94 |
+
"use_sliding_window": false,
|
95 |
+
"vocab_size": 151936
|
96 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/Qwen/Qwen2.5-0.5B/8a714805c0774dccda3d.json
ADDED
@@ -0,0 +1,96 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "Qwen/Qwen2.5-0.5B",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen2ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 896,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 4864,
|
13 |
+
"layer_types": [
|
14 |
+
"full_attention",
|
15 |
+
"full_attention",
|
16 |
+
"full_attention",
|
17 |
+
"full_attention",
|
18 |
+
"full_attention",
|
19 |
+
"full_attention",
|
20 |
+
"full_attention",
|
21 |
+
"full_attention",
|
22 |
+
"full_attention",
|
23 |
+
"full_attention",
|
24 |
+
"full_attention",
|
25 |
+
"full_attention",
|
26 |
+
"full_attention",
|
27 |
+
"full_attention",
|
28 |
+
"full_attention",
|
29 |
+
"full_attention",
|
30 |
+
"full_attention",
|
31 |
+
"full_attention",
|
32 |
+
"full_attention",
|
33 |
+
"full_attention",
|
34 |
+
"full_attention",
|
35 |
+
"full_attention",
|
36 |
+
"full_attention",
|
37 |
+
"full_attention"
|
38 |
+
],
|
39 |
+
"max_position_embeddings": 32768,
|
40 |
+
"max_window_layers": 24,
|
41 |
+
"model_type": "qwen2",
|
42 |
+
"neuron": {
|
43 |
+
"_serialized_key": "NxDNeuronConfig",
|
44 |
+
"async_mode": false,
|
45 |
+
"attn_kernel_enabled": false,
|
46 |
+
"batch_size": 1,
|
47 |
+
"capacity_factor": null,
|
48 |
+
"cc_pipeline_tiling_factor": 2,
|
49 |
+
"checkpoint_id": "Qwen/Qwen2.5-0.5B",
|
50 |
+
"checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
|
51 |
+
"continuous_batching": false,
|
52 |
+
"enable_bucketing": false,
|
53 |
+
"ep_degree": 1,
|
54 |
+
"flash_decoding_enabled": false,
|
55 |
+
"fused_qkv": false,
|
56 |
+
"glu_mlp": true,
|
57 |
+
"is_chunked_prefill": false,
|
58 |
+
"local_ranks_size": 2,
|
59 |
+
"logical_nc_config": 1,
|
60 |
+
"max_batch_size": 1,
|
61 |
+
"max_context_length": 32768,
|
62 |
+
"max_topk": 256,
|
63 |
+
"mlp_kernel_enabled": false,
|
64 |
+
"mlp_kernel_fuse_residual_add": false,
|
65 |
+
"n_active_tokens": 32768,
|
66 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
67 |
+
"num_cores_per_group": 1,
|
68 |
+
"on_device_sampling": true,
|
69 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
70 |
+
"output_logits": false,
|
71 |
+
"padding_side": "right",
|
72 |
+
"pp_degree": 1,
|
73 |
+
"qkv_kernel_enabled": false,
|
74 |
+
"rpl_reduce_dtype": "bfloat16",
|
75 |
+
"sequence_length": 32768,
|
76 |
+
"sequence_parallel_enabled": false,
|
77 |
+
"speculation_length": 0,
|
78 |
+
"start_rank_id": 0,
|
79 |
+
"target": null,
|
80 |
+
"torch_dtype": "bfloat16",
|
81 |
+
"tp_degree": 2,
|
82 |
+
"vocab_parallel": false
|
83 |
+
},
|
84 |
+
"num_attention_heads": 14,
|
85 |
+
"num_hidden_layers": 24,
|
86 |
+
"num_key_value_heads": 2,
|
87 |
+
"rms_norm_eps": 1e-06,
|
88 |
+
"rope_scaling": null,
|
89 |
+
"rope_theta": 1000000.0,
|
90 |
+
"sliding_window": null,
|
91 |
+
"tie_word_embeddings": true,
|
92 |
+
"use_cache": true,
|
93 |
+
"use_mrope": false,
|
94 |
+
"use_sliding_window": false,
|
95 |
+
"vocab_size": 151936
|
96 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/yujiepan/qwen2.5-128k-tiny-random/4be7e684c15f704cb67d.json
ADDED
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/qwen2.5-128k-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen2ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 8,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 16,
|
13 |
+
"layer_types": [
|
14 |
+
"full_attention",
|
15 |
+
"full_attention"
|
16 |
+
],
|
17 |
+
"max_position_embeddings": 32768,
|
18 |
+
"max_window_layers": 1,
|
19 |
+
"model_type": "qwen2",
|
20 |
+
"neuron": {
|
21 |
+
"_serialized_key": "NxDNeuronConfig",
|
22 |
+
"async_mode": false,
|
23 |
+
"attn_kernel_enabled": false,
|
24 |
+
"batch_size": 2,
|
25 |
+
"capacity_factor": null,
|
26 |
+
"cc_pipeline_tiling_factor": 2,
|
27 |
+
"checkpoint_id": "yujiepan/qwen2.5-128k-tiny-random",
|
28 |
+
"checkpoint_revision": "c8296d4ca3f87782876d2382fbb6481d1beb8ef0",
|
29 |
+
"continuous_batching": true,
|
30 |
+
"enable_bucketing": false,
|
31 |
+
"ep_degree": 1,
|
32 |
+
"flash_decoding_enabled": false,
|
33 |
+
"fused_qkv": false,
|
34 |
+
"glu_mlp": true,
|
35 |
+
"is_chunked_prefill": false,
|
36 |
+
"local_ranks_size": 2,
|
37 |
+
"logical_nc_config": 1,
|
38 |
+
"max_batch_size": 2,
|
39 |
+
"max_context_length": 100,
|
40 |
+
"max_topk": 256,
|
41 |
+
"mlp_kernel_enabled": false,
|
42 |
+
"mlp_kernel_fuse_residual_add": false,
|
43 |
+
"n_active_tokens": 100,
|
44 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
45 |
+
"num_cores_per_group": 1,
|
46 |
+
"on_device_sampling": false,
|
47 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
48 |
+
"output_logits": false,
|
49 |
+
"padding_side": "right",
|
50 |
+
"pp_degree": 1,
|
51 |
+
"qkv_kernel_enabled": false,
|
52 |
+
"rpl_reduce_dtype": "float16",
|
53 |
+
"sequence_length": 100,
|
54 |
+
"sequence_parallel_enabled": false,
|
55 |
+
"speculation_length": 0,
|
56 |
+
"start_rank_id": 0,
|
57 |
+
"target": null,
|
58 |
+
"torch_dtype": "float16",
|
59 |
+
"tp_degree": 2,
|
60 |
+
"vocab_parallel": false
|
61 |
+
},
|
62 |
+
"num_attention_heads": 4,
|
63 |
+
"num_hidden_layers": 2,
|
64 |
+
"num_key_value_heads": 2,
|
65 |
+
"rms_norm_eps": 1e-06,
|
66 |
+
"rope_scaling": {
|
67 |
+
"factor": 4.0,
|
68 |
+
"original_max_position_embeddings": 32768,
|
69 |
+
"rope_type": "yarn",
|
70 |
+
"type": "yarn"
|
71 |
+
},
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"sliding_window": null,
|
74 |
+
"tie_word_embeddings": false,
|
75 |
+
"use_cache": true,
|
76 |
+
"use_sliding_window": false,
|
77 |
+
"vocab_size": 152064
|
78 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/yujiepan/qwen2.5-128k-tiny-random/6cead920a0dbb3daefb9.json
ADDED
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/qwen2.5-128k-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen2ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 8,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 16,
|
13 |
+
"layer_types": [
|
14 |
+
"full_attention",
|
15 |
+
"full_attention"
|
16 |
+
],
|
17 |
+
"max_position_embeddings": 32768,
|
18 |
+
"max_window_layers": 1,
|
19 |
+
"model_type": "qwen2",
|
20 |
+
"neuron": {
|
21 |
+
"_serialized_key": "NxDNeuronConfig",
|
22 |
+
"async_mode": false,
|
23 |
+
"attn_kernel_enabled": false,
|
24 |
+
"batch_size": 1,
|
25 |
+
"capacity_factor": null,
|
26 |
+
"cc_pipeline_tiling_factor": 2,
|
27 |
+
"checkpoint_id": "yujiepan/qwen2.5-128k-tiny-random",
|
28 |
+
"checkpoint_revision": "c8296d4ca3f87782876d2382fbb6481d1beb8ef0",
|
29 |
+
"continuous_batching": false,
|
30 |
+
"enable_bucketing": false,
|
31 |
+
"ep_degree": 1,
|
32 |
+
"flash_decoding_enabled": false,
|
33 |
+
"fused_qkv": false,
|
34 |
+
"glu_mlp": true,
|
35 |
+
"is_chunked_prefill": false,
|
36 |
+
"local_ranks_size": 2,
|
37 |
+
"logical_nc_config": 1,
|
38 |
+
"max_batch_size": 1,
|
39 |
+
"max_context_length": 100,
|
40 |
+
"max_topk": 256,
|
41 |
+
"mlp_kernel_enabled": false,
|
42 |
+
"mlp_kernel_fuse_residual_add": false,
|
43 |
+
"n_active_tokens": 100,
|
44 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
45 |
+
"num_cores_per_group": 1,
|
46 |
+
"on_device_sampling": true,
|
47 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
48 |
+
"output_logits": false,
|
49 |
+
"padding_side": "right",
|
50 |
+
"pp_degree": 1,
|
51 |
+
"qkv_kernel_enabled": false,
|
52 |
+
"rpl_reduce_dtype": "float16",
|
53 |
+
"sequence_length": 100,
|
54 |
+
"sequence_parallel_enabled": false,
|
55 |
+
"speculation_length": 0,
|
56 |
+
"start_rank_id": 0,
|
57 |
+
"target": null,
|
58 |
+
"torch_dtype": "float16",
|
59 |
+
"tp_degree": 2,
|
60 |
+
"vocab_parallel": false
|
61 |
+
},
|
62 |
+
"num_attention_heads": 4,
|
63 |
+
"num_hidden_layers": 2,
|
64 |
+
"num_key_value_heads": 2,
|
65 |
+
"rms_norm_eps": 1e-06,
|
66 |
+
"rope_scaling": {
|
67 |
+
"factor": 4.0,
|
68 |
+
"original_max_position_embeddings": 32768,
|
69 |
+
"rope_type": "yarn",
|
70 |
+
"type": "yarn"
|
71 |
+
},
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"sliding_window": null,
|
74 |
+
"tie_word_embeddings": false,
|
75 |
+
"use_cache": true,
|
76 |
+
"use_sliding_window": false,
|
77 |
+
"vocab_size": 152064
|
78 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen2/yujiepan/qwen2.5-128k-tiny-random/cd761b198c1bec21bd55.json
ADDED
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "yujiepan/qwen2.5-128k-tiny-random",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen2ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 8,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 16,
|
13 |
+
"layer_types": [
|
14 |
+
"full_attention",
|
15 |
+
"full_attention"
|
16 |
+
],
|
17 |
+
"max_position_embeddings": 32768,
|
18 |
+
"max_window_layers": 1,
|
19 |
+
"model_type": "qwen2",
|
20 |
+
"neuron": {
|
21 |
+
"_serialized_key": "NxDNeuronConfig",
|
22 |
+
"async_mode": false,
|
23 |
+
"attn_kernel_enabled": false,
|
24 |
+
"batch_size": 1,
|
25 |
+
"capacity_factor": null,
|
26 |
+
"cc_pipeline_tiling_factor": 2,
|
27 |
+
"checkpoint_id": "yujiepan/qwen2.5-128k-tiny-random",
|
28 |
+
"checkpoint_revision": "c8296d4ca3f87782876d2382fbb6481d1beb8ef0",
|
29 |
+
"continuous_batching": false,
|
30 |
+
"enable_bucketing": false,
|
31 |
+
"ep_degree": 1,
|
32 |
+
"flash_decoding_enabled": false,
|
33 |
+
"fused_qkv": false,
|
34 |
+
"glu_mlp": true,
|
35 |
+
"is_chunked_prefill": false,
|
36 |
+
"local_ranks_size": 2,
|
37 |
+
"logical_nc_config": 1,
|
38 |
+
"max_batch_size": 1,
|
39 |
+
"max_context_length": 100,
|
40 |
+
"max_topk": 256,
|
41 |
+
"mlp_kernel_enabled": false,
|
42 |
+
"mlp_kernel_fuse_residual_add": false,
|
43 |
+
"n_active_tokens": 100,
|
44 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
45 |
+
"num_cores_per_group": 1,
|
46 |
+
"on_device_sampling": true,
|
47 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
48 |
+
"output_logits": false,
|
49 |
+
"padding_side": "right",
|
50 |
+
"pp_degree": 1,
|
51 |
+
"qkv_kernel_enabled": false,
|
52 |
+
"rpl_reduce_dtype": "bfloat16",
|
53 |
+
"sequence_length": 100,
|
54 |
+
"sequence_parallel_enabled": false,
|
55 |
+
"speculation_length": 0,
|
56 |
+
"start_rank_id": 0,
|
57 |
+
"target": null,
|
58 |
+
"torch_dtype": "bfloat16",
|
59 |
+
"tp_degree": 2,
|
60 |
+
"vocab_parallel": false
|
61 |
+
},
|
62 |
+
"num_attention_heads": 4,
|
63 |
+
"num_hidden_layers": 2,
|
64 |
+
"num_key_value_heads": 2,
|
65 |
+
"rms_norm_eps": 1e-06,
|
66 |
+
"rope_scaling": {
|
67 |
+
"factor": 4.0,
|
68 |
+
"original_max_position_embeddings": 32768,
|
69 |
+
"rope_type": "yarn",
|
70 |
+
"type": "yarn"
|
71 |
+
},
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"sliding_window": null,
|
74 |
+
"tie_word_embeddings": false,
|
75 |
+
"use_cache": true,
|
76 |
+
"use_sliding_window": false,
|
77 |
+
"vocab_size": 152064
|
78 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3/Qwen/Qwen3-0.6B/2ed7a8812dbe9c7ab058.json
ADDED
@@ -0,0 +1,101 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "Qwen/Qwen3-0.6B",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"head_dim": 128,
|
11 |
+
"hidden_act": "silu",
|
12 |
+
"hidden_size": 1024,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 3072,
|
15 |
+
"layer_types": [
|
16 |
+
"full_attention",
|
17 |
+
"full_attention",
|
18 |
+
"full_attention",
|
19 |
+
"full_attention",
|
20 |
+
"full_attention",
|
21 |
+
"full_attention",
|
22 |
+
"full_attention",
|
23 |
+
"full_attention",
|
24 |
+
"full_attention",
|
25 |
+
"full_attention",
|
26 |
+
"full_attention",
|
27 |
+
"full_attention",
|
28 |
+
"full_attention",
|
29 |
+
"full_attention",
|
30 |
+
"full_attention",
|
31 |
+
"full_attention",
|
32 |
+
"full_attention",
|
33 |
+
"full_attention",
|
34 |
+
"full_attention",
|
35 |
+
"full_attention",
|
36 |
+
"full_attention",
|
37 |
+
"full_attention",
|
38 |
+
"full_attention",
|
39 |
+
"full_attention",
|
40 |
+
"full_attention",
|
41 |
+
"full_attention",
|
42 |
+
"full_attention",
|
43 |
+
"full_attention"
|
44 |
+
],
|
45 |
+
"max_position_embeddings": 40960,
|
46 |
+
"max_window_layers": 28,
|
47 |
+
"model_type": "qwen3",
|
48 |
+
"neuron": {
|
49 |
+
"_serialized_key": "NxDNeuronConfig",
|
50 |
+
"async_mode": false,
|
51 |
+
"attn_kernel_enabled": false,
|
52 |
+
"batch_size": 4,
|
53 |
+
"capacity_factor": null,
|
54 |
+
"cc_pipeline_tiling_factor": 2,
|
55 |
+
"checkpoint_id": "Qwen/Qwen3-0.6B",
|
56 |
+
"checkpoint_revision": "c1899de289a04d12100db370d81485cdf75e47ca",
|
57 |
+
"continuous_batching": true,
|
58 |
+
"enable_bucketing": false,
|
59 |
+
"ep_degree": 1,
|
60 |
+
"flash_decoding_enabled": false,
|
61 |
+
"fused_qkv": true,
|
62 |
+
"glu_mlp": true,
|
63 |
+
"is_chunked_prefill": false,
|
64 |
+
"local_ranks_size": 2,
|
65 |
+
"logical_nc_config": 1,
|
66 |
+
"max_batch_size": 4,
|
67 |
+
"max_context_length": 4096,
|
68 |
+
"max_topk": 256,
|
69 |
+
"mlp_kernel_enabled": false,
|
70 |
+
"mlp_kernel_fuse_residual_add": false,
|
71 |
+
"n_active_tokens": 4096,
|
72 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
73 |
+
"num_cores_per_group": 1,
|
74 |
+
"on_device_sampling": false,
|
75 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
76 |
+
"output_logits": false,
|
77 |
+
"padding_side": "right",
|
78 |
+
"pp_degree": 1,
|
79 |
+
"qkv_kernel_enabled": false,
|
80 |
+
"rpl_reduce_dtype": "bfloat16",
|
81 |
+
"sequence_length": 4096,
|
82 |
+
"sequence_parallel_enabled": false,
|
83 |
+
"speculation_length": 0,
|
84 |
+
"start_rank_id": 0,
|
85 |
+
"target": null,
|
86 |
+
"torch_dtype": "bfloat16",
|
87 |
+
"tp_degree": 2,
|
88 |
+
"vocab_parallel": false
|
89 |
+
},
|
90 |
+
"num_attention_heads": 16,
|
91 |
+
"num_hidden_layers": 28,
|
92 |
+
"num_key_value_heads": 8,
|
93 |
+
"rms_norm_eps": 1e-06,
|
94 |
+
"rope_scaling": null,
|
95 |
+
"rope_theta": 1000000,
|
96 |
+
"sliding_window": null,
|
97 |
+
"tie_word_embeddings": true,
|
98 |
+
"use_cache": true,
|
99 |
+
"use_sliding_window": false,
|
100 |
+
"vocab_size": 151936
|
101 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/8fff447965fd3c157c0f.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 2,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
29 |
+
"checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 2,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/90f9e02faff8566070cb.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
29 |
+
"checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "float16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "float16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/f6c02365716b13dea692.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 2,
|
11 |
+
"head_dim": 32,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 64,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 128,
|
16 |
+
"max_position_embeddings": 40960,
|
17 |
+
"max_window_layers": 1,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 128,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 1,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
|
29 |
+
"checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 2,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 1,
|
40 |
+
"max_context_length": 100,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 100,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 100,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 2,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 2,
|
65 |
+
"num_experts": 8,
|
66 |
+
"num_experts_per_tok": 2,
|
67 |
+
"num_hidden_layers": 2,
|
68 |
+
"num_key_value_heads": 1,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 1000000.0,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": true,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev2/smollm3/HuggingFaceTB/SmolLM3-3B/60de13af0adf5a679b2c.json
ADDED
@@ -0,0 +1,148 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "HuggingFaceTB/SmolLM3-3B",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"SmolLM3ForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"hidden_act": "silu",
|
11 |
+
"hidden_size": 2048,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 11008,
|
14 |
+
"layer_types": [
|
15 |
+
"full_attention",
|
16 |
+
"full_attention",
|
17 |
+
"full_attention",
|
18 |
+
"full_attention",
|
19 |
+
"full_attention",
|
20 |
+
"full_attention",
|
21 |
+
"full_attention",
|
22 |
+
"full_attention",
|
23 |
+
"full_attention",
|
24 |
+
"full_attention",
|
25 |
+
"full_attention",
|
26 |
+
"full_attention",
|
27 |
+
"full_attention",
|
28 |
+
"full_attention",
|
29 |
+
"full_attention",
|
30 |
+
"full_attention",
|
31 |
+
"full_attention",
|
32 |
+
"full_attention",
|
33 |
+
"full_attention",
|
34 |
+
"full_attention",
|
35 |
+
"full_attention",
|
36 |
+
"full_attention",
|
37 |
+
"full_attention",
|
38 |
+
"full_attention",
|
39 |
+
"full_attention",
|
40 |
+
"full_attention",
|
41 |
+
"full_attention",
|
42 |
+
"full_attention",
|
43 |
+
"full_attention",
|
44 |
+
"full_attention",
|
45 |
+
"full_attention",
|
46 |
+
"full_attention",
|
47 |
+
"full_attention",
|
48 |
+
"full_attention",
|
49 |
+
"full_attention",
|
50 |
+
"full_attention"
|
51 |
+
],
|
52 |
+
"max_position_embeddings": 65536,
|
53 |
+
"max_window_layers": 28,
|
54 |
+
"mlp_bias": false,
|
55 |
+
"model_type": "smollm3",
|
56 |
+
"neuron": {
|
57 |
+
"_serialized_key": "NxDNeuronConfig",
|
58 |
+
"async_mode": false,
|
59 |
+
"attn_kernel_enabled": false,
|
60 |
+
"batch_size": 4,
|
61 |
+
"capacity_factor": null,
|
62 |
+
"cc_pipeline_tiling_factor": 2,
|
63 |
+
"checkpoint_id": "HuggingFaceTB/SmolLM3-3B",
|
64 |
+
"checkpoint_revision": "1c00fc78bd9cf90108046bc433cb34992480f1c1",
|
65 |
+
"continuous_batching": true,
|
66 |
+
"enable_bucketing": false,
|
67 |
+
"ep_degree": 1,
|
68 |
+
"flash_decoding_enabled": false,
|
69 |
+
"fused_qkv": true,
|
70 |
+
"glu_mlp": true,
|
71 |
+
"is_chunked_prefill": false,
|
72 |
+
"local_ranks_size": 2,
|
73 |
+
"logical_nc_config": 1,
|
74 |
+
"max_batch_size": 4,
|
75 |
+
"max_context_length": 4096,
|
76 |
+
"max_topk": 256,
|
77 |
+
"mlp_kernel_enabled": false,
|
78 |
+
"mlp_kernel_fuse_residual_add": false,
|
79 |
+
"n_active_tokens": 4096,
|
80 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
81 |
+
"num_cores_per_group": 1,
|
82 |
+
"on_device_sampling": true,
|
83 |
+
"optimum_neuron_version": "0.3.1.dev2",
|
84 |
+
"output_logits": false,
|
85 |
+
"padding_side": "right",
|
86 |
+
"pp_degree": 1,
|
87 |
+
"qkv_kernel_enabled": false,
|
88 |
+
"rpl_reduce_dtype": "bfloat16",
|
89 |
+
"sequence_length": 4096,
|
90 |
+
"sequence_parallel_enabled": false,
|
91 |
+
"speculation_length": 0,
|
92 |
+
"start_rank_id": 0,
|
93 |
+
"target": null,
|
94 |
+
"torch_dtype": "bfloat16",
|
95 |
+
"tp_degree": 2,
|
96 |
+
"vocab_parallel": false
|
97 |
+
},
|
98 |
+
"no_rope_layer_interval": 4,
|
99 |
+
"no_rope_layers": [
|
100 |
+
1,
|
101 |
+
1,
|
102 |
+
1,
|
103 |
+
0,
|
104 |
+
1,
|
105 |
+
1,
|
106 |
+
1,
|
107 |
+
0,
|
108 |
+
1,
|
109 |
+
1,
|
110 |
+
1,
|
111 |
+
0,
|
112 |
+
1,
|
113 |
+
1,
|
114 |
+
1,
|
115 |
+
0,
|
116 |
+
1,
|
117 |
+
1,
|
118 |
+
1,
|
119 |
+
0,
|
120 |
+
1,
|
121 |
+
1,
|
122 |
+
1,
|
123 |
+
0,
|
124 |
+
1,
|
125 |
+
1,
|
126 |
+
1,
|
127 |
+
0,
|
128 |
+
1,
|
129 |
+
1,
|
130 |
+
1,
|
131 |
+
0,
|
132 |
+
1,
|
133 |
+
1,
|
134 |
+
1,
|
135 |
+
0
|
136 |
+
],
|
137 |
+
"num_attention_heads": 16,
|
138 |
+
"num_hidden_layers": 36,
|
139 |
+
"num_key_value_heads": 4,
|
140 |
+
"pretraining_tp": 2,
|
141 |
+
"rms_norm_eps": 1e-06,
|
142 |
+
"rope_scaling": null,
|
143 |
+
"rope_theta": 5000000.0,
|
144 |
+
"sliding_window": null,
|
145 |
+
"use_cache": false,
|
146 |
+
"use_sliding_window": false,
|
147 |
+
"vocab_size": 128256
|
148 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/0_REGISTRY/0.3.1.dev3/qwen3_moe/Qwen/Qwen3-30B-A3B-Instruct-2507/fc72a21b6de27e9bcefe.json
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_entry_class": "SingleModelCacheEntry",
|
3 |
+
"_model_id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
|
4 |
+
"_task": "text-generation",
|
5 |
+
"architectures": [
|
6 |
+
"Qwen3MoeForCausalLM"
|
7 |
+
],
|
8 |
+
"attention_bias": false,
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"decoder_sparse_step": 1,
|
11 |
+
"head_dim": 128,
|
12 |
+
"hidden_act": "silu",
|
13 |
+
"hidden_size": 2048,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 6144,
|
16 |
+
"max_position_embeddings": 262144,
|
17 |
+
"max_window_layers": 48,
|
18 |
+
"mlp_only_layers": [],
|
19 |
+
"model_type": "qwen3_moe",
|
20 |
+
"moe_intermediate_size": 768,
|
21 |
+
"neuron": {
|
22 |
+
"_serialized_key": "NxDNeuronConfig",
|
23 |
+
"async_mode": false,
|
24 |
+
"attn_kernel_enabled": false,
|
25 |
+
"batch_size": 8,
|
26 |
+
"capacity_factor": null,
|
27 |
+
"cc_pipeline_tiling_factor": 2,
|
28 |
+
"checkpoint_id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
|
29 |
+
"checkpoint_revision": "61082d4deaa4785f64943b443cbc2b5de7524fad",
|
30 |
+
"continuous_batching": false,
|
31 |
+
"enable_bucketing": false,
|
32 |
+
"ep_degree": 1,
|
33 |
+
"flash_decoding_enabled": false,
|
34 |
+
"fused_qkv": false,
|
35 |
+
"glu_mlp": true,
|
36 |
+
"is_chunked_prefill": false,
|
37 |
+
"local_ranks_size": 8,
|
38 |
+
"logical_nc_config": 1,
|
39 |
+
"max_batch_size": 8,
|
40 |
+
"max_context_length": 4096,
|
41 |
+
"max_topk": 256,
|
42 |
+
"mlp_kernel_enabled": false,
|
43 |
+
"mlp_kernel_fuse_residual_add": false,
|
44 |
+
"n_active_tokens": 4096,
|
45 |
+
"neuronxcc_version": "2.19.8089.0+8ab9f450",
|
46 |
+
"num_cores_per_group": 1,
|
47 |
+
"on_device_sampling": false,
|
48 |
+
"optimum_neuron_version": "0.3.1.dev3",
|
49 |
+
"output_logits": false,
|
50 |
+
"padding_side": "right",
|
51 |
+
"pp_degree": 1,
|
52 |
+
"qkv_kernel_enabled": false,
|
53 |
+
"rpl_reduce_dtype": "bfloat16",
|
54 |
+
"sequence_length": 4096,
|
55 |
+
"sequence_parallel_enabled": false,
|
56 |
+
"speculation_length": 0,
|
57 |
+
"start_rank_id": 0,
|
58 |
+
"target": null,
|
59 |
+
"torch_dtype": "bfloat16",
|
60 |
+
"tp_degree": 8,
|
61 |
+
"vocab_parallel": false
|
62 |
+
},
|
63 |
+
"norm_topk_prob": true,
|
64 |
+
"num_attention_heads": 32,
|
65 |
+
"num_experts": 128,
|
66 |
+
"num_experts_per_tok": 8,
|
67 |
+
"num_hidden_layers": 48,
|
68 |
+
"num_key_value_heads": 4,
|
69 |
+
"output_router_logits": false,
|
70 |
+
"rms_norm_eps": 1e-06,
|
71 |
+
"rope_scaling": null,
|
72 |
+
"rope_theta": 10000000,
|
73 |
+
"router_aux_loss_coef": 0.001,
|
74 |
+
"sliding_window": null,
|
75 |
+
"tie_word_embeddings": false,
|
76 |
+
"use_cache": true,
|
77 |
+
"use_sliding_window": false,
|
78 |
+
"vocab_size": 151936
|
79 |
+
}
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/compile_flags.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
"--model-type=transformer -O1 --lnc=1 --internal-hlo2tensorizer-options=--experimental-unsafe-fp8e4m3fn-as-fp8e4m3 --logfile=/tmp/nxd_model/layout_opt/log-neuron-cc.txt"
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/model.done
ADDED
File without changes
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/model.hlo_module.pb
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1a6a5e05e7901f7ec6ea776377316585b1bf917c33884f8a2a2d3851e28128c6
|
3 |
+
size 7993
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_04376e1307045cf435ae+431f5505/model.neff
ADDED
Binary file (42 kB). View file
|
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/compile_flags.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
["--target=trn1", "--enable-saturate-infinity", "--enable-mixed-precision-accumulation", "--model-type", "transformer", "-O1", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2", "--auto-cast=none", "--internal-enable-dge-levels", "vector_dynamic_offsets", "--internal-hlo2tensorizer-options=--verify-hlo=true", "--logfile=/tmp/nxd_model/token_generation_model/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/model.done
ADDED
File without changes
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/model.hlo_module.pb
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a31c4cdb1fbc97738a6e37dec33d6f44e6b1284cef02f7149efa6cfbe91b480c
|
3 |
+
size 63836
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/model.neff
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8842daaea9ad716443a64507d5e6eb80bd5036701c69d3cc22d3c8b28cfe3129
|
3 |
+
size 297984
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0a47deab436eaf26c99a+cd3419b6/wrapped_neff.hlo
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:01904b235c4fe11e10878eb7201accabe15e04a5a33cb5e989bab9b2b9fd2617
|
3 |
+
size 307570
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0b67a734fc55d67768bd+253d6470/compile_flags.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
["--target=trn1", "--enable-saturate-infinity", "--enable-mixed-precision-accumulation", "--model-type", "transformer", "-O1", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2", "--auto-cast=none", "--internal-enable-dge-levels", "vector_dynamic_offsets", "--internal-hlo2tensorizer-options=--verify-hlo=true", "--logfile=/tmp/nxd_model/context_encoding_model/_tp0_bk0/log-neuron-cc.txt"]
|
neuronxcc-2.19.8089.0+8ab9f450/MODULE_0b67a734fc55d67768bd+253d6470/model.done
ADDED
File without changes
|