Upload config.json
Browse files- config.json +36 -0
config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/project2/lichanghao_w/lch/retrain-heyna/checkpoint-15000",
|
| 3 |
+
"activation_freq": 10,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"PlantGLMForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"auto_map": {
|
| 8 |
+
"AutoConfig": "configuration_plantglm.PlantGLMConfig",
|
| 9 |
+
"AutoModel": "modeling_plantglm.PlantGLMModel",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_plantglm.PlantGLMForCausalLM",
|
| 11 |
+
"AutoModelForSequenceClassification": "modeling_plantglm.PlantGLMForSequenceClassification"
|
| 12 |
+
},
|
| 13 |
+
"d_inner": 3072,
|
| 14 |
+
"d_model": 1024,
|
| 15 |
+
"emb_dim": 9,
|
| 16 |
+
"embed_dropout": 0.1,
|
| 17 |
+
"filter_order": 256,
|
| 18 |
+
"hyena_dropout": 0.0,
|
| 19 |
+
"hyena_filter_dropout": 0.0,
|
| 20 |
+
"hyena_order": 2,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"max_seq_len": 65538,
|
| 23 |
+
"model_type": "plantglm",
|
| 24 |
+
"n_layer": 16,
|
| 25 |
+
"num_inner_mlps": 2,
|
| 26 |
+
"pad_token_id": 0,
|
| 27 |
+
"pad_vocab_size_multiple": 8,
|
| 28 |
+
"rms_norm_epsilon": 1e-06,
|
| 29 |
+
"short_filter_order": 3,
|
| 30 |
+
"tie_word_embeddings": false,
|
| 31 |
+
"torch_dtype": "float32",
|
| 32 |
+
"train_freq": true,
|
| 33 |
+
"transformers_version": "4.44.2",
|
| 34 |
+
"use_bias": true,
|
| 35 |
+
"vocab_size": 24
|
| 36 |
+
}
|