Create config.json
Browse files- config.json +26 -0
config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "gelu",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LNNModel"
|
| 5 |
+
],
|
| 6 |
+
"dt": 0.1,
|
| 7 |
+
"eos_token_id": 1,
|
| 8 |
+
"expert_dim": 32768,
|
| 9 |
+
"hidden_size": 3048,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"lambda_res": 0.0,
|
| 12 |
+
"model_type": "quasar",
|
| 13 |
+
"moe_load_balance_loss_weight": 0.01,
|
| 14 |
+
"num_experts": 14,
|
| 15 |
+
"num_experts_per_tok": 2,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 1,
|
| 18 |
+
"pmb_num_blocks": 1024,
|
| 19 |
+
"pmb_slots_per_block": 4096,
|
| 20 |
+
"pmb_top_k": 1,
|
| 21 |
+
"torch_dtype": "float32",
|
| 22 |
+
"transformers_version": "4.35.2",
|
| 23 |
+
"use_moe": true,
|
| 24 |
+
"use_pmb": false,
|
| 25 |
+
"vocab_size": 128815
|
| 26 |
+
}
|