Commit
•
08601ed
1
Parent(s):
98dfc1e
增加llama3-8b
Browse files
Meta-Llama-3-8B-Instruct/bert4torch_config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "llama",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"intermediate_size": 14336,
|
| 5 |
+
"num_attention_heads": 32,
|
| 6 |
+
"num_hidden_layers": 32,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"vocab_size": 128256,
|
| 9 |
+
"segment_vocab_size": 0,
|
| 10 |
+
"skip_init": true,
|
| 11 |
+
"layer_norm_eps": 1e-05,
|
| 12 |
+
"rope_rank": "updown",
|
| 13 |
+
"rope_theta": 500000.0,
|
| 14 |
+
"num_key_value_heads": 8,
|
| 15 |
+
"max_position_embeddings": 8192,
|
| 16 |
+
"torch_dtype": "bfloat16",
|
| 17 |
+
"tie_word_embeddings": false,
|
| 18 |
+
"attention_probs_dropout_prob": 0.0
|
| 19 |
+
}
|
Meta-Llama-3-8B/bert4torch_config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "llama",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"intermediate_size": 14336,
|
| 5 |
+
"num_attention_heads": 32,
|
| 6 |
+
"num_hidden_layers": 32,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"vocab_size": 128256,
|
| 9 |
+
"segment_vocab_size": 0,
|
| 10 |
+
"skip_init": true,
|
| 11 |
+
"layer_norm_eps": 1e-05,
|
| 12 |
+
"rope_rank": "updown",
|
| 13 |
+
"rope_theta": 500000.0,
|
| 14 |
+
"num_key_value_heads": 8,
|
| 15 |
+
"max_position_embeddings": 8192,
|
| 16 |
+
"torch_dtype": "bfloat16",
|
| 17 |
+
"tie_word_embeddings": false,
|
| 18 |
+
"attention_probs_dropout_prob": 0.0
|
| 19 |
+
}
|