First model version
Browse files- config.json +38 -0
config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"OpenMoeForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"capacity_factor_eval": 2.0,
|
| 7 |
+
"capacity_factor_train": 1.25,
|
| 8 |
+
"drop_tks": true,
|
| 9 |
+
"dropout_rate": 0.0,
|
| 10 |
+
"eos_token_id": 2,
|
| 11 |
+
"expert_parallel": null,
|
| 12 |
+
"gated": true,
|
| 13 |
+
"head_dim": 128,
|
| 14 |
+
"hidden_act": "swiglu",
|
| 15 |
+
"hidden_size": 2048,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 8192,
|
| 18 |
+
"layer_norm_epsilon": 1e-06,
|
| 19 |
+
"max_position_embeddings": 2048,
|
| 20 |
+
"min_capacity": 4,
|
| 21 |
+
"model_type": "llama",
|
| 22 |
+
"moe_layer_interval": 6,
|
| 23 |
+
"noisy_policy": null,
|
| 24 |
+
"num_attention_heads": 24,
|
| 25 |
+
"num_experts": 32,
|
| 26 |
+
"num_hidden_layers": 24,
|
| 27 |
+
"num_key_value_heads": 24,
|
| 28 |
+
"pretraining_tp": 1,
|
| 29 |
+
"rms_norm_eps": 1e-06,
|
| 30 |
+
"rope_scaling": null,
|
| 31 |
+
"rope_theta": 10000.0,
|
| 32 |
+
"tie_word_embeddings": false,
|
| 33 |
+
"topk": 2,
|
| 34 |
+
"torch_dtype": "float32",
|
| 35 |
+
"transformers_version": "4.33.1",
|
| 36 |
+
"use_cache": true,
|
| 37 |
+
"vocab_size": 256384
|
| 38 |
+
}
|