v2-epoch-2: v2 config
Browse files- v2-epoch-2/config.json +26 -0
v2-epoch-2/config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"model_type": "llama",
|
| 6 |
+
"hidden_size": 3072,
|
| 7 |
+
"intermediate_size": 8192,
|
| 8 |
+
"num_hidden_layers": 4,
|
| 9 |
+
"num_attention_heads": 24,
|
| 10 |
+
"num_key_value_heads": 8,
|
| 11 |
+
"head_dim": 128,
|
| 12 |
+
"hidden_act": "silu",
|
| 13 |
+
"max_position_embeddings": 196608,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"rms_norm_eps": 1e-06,
|
| 16 |
+
"rope_theta": 10000.0,
|
| 17 |
+
"vocab_size": 200064,
|
| 18 |
+
"draft_vocab_size": 200064,
|
| 19 |
+
"attention_bias": false,
|
| 20 |
+
"attention_dropout": 0.0,
|
| 21 |
+
"mlp_bias": false,
|
| 22 |
+
"tie_word_embeddings": false,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"dtype": "bfloat16",
|
| 25 |
+
"ptd_token_id": 200063
|
| 26 |
+
}
|