Delta-Vector
commited on
Commit
•
0bf1b89
1
Parent(s):
f60c226
Update config.json
Browse files- config.json +1 -8
config.json
CHANGED
@@ -12,7 +12,7 @@
|
|
12 |
"hidden_size": 3072,
|
13 |
"initializer_range": 0.02,
|
14 |
"intermediate_size": 9216,
|
15 |
-
"max_position_embeddings":
|
16 |
"mlp_bias": false,
|
17 |
"model_type": "llama",
|
18 |
"num_attention_heads": 32,
|
@@ -20,13 +20,6 @@
|
|
20 |
"num_key_value_heads": 8,
|
21 |
"pretraining_tp": 1,
|
22 |
"rms_norm_eps": 1e-05,
|
23 |
-
"rope_scaling": {
|
24 |
-
"factor": 8.0,
|
25 |
-
"high_freq_factor": 4.0,
|
26 |
-
"low_freq_factor": 1.0,
|
27 |
-
"original_max_position_embeddings": 8192,
|
28 |
-
"rope_type": "llama3"
|
29 |
-
},
|
30 |
"rope_theta": 500000.0,
|
31 |
"tie_word_embeddings": false,
|
32 |
"torch_dtype": "bfloat16",
|
|
|
12 |
"hidden_size": 3072,
|
13 |
"initializer_range": 0.02,
|
14 |
"intermediate_size": 9216,
|
15 |
+
"max_position_embeddings": 8192,
|
16 |
"mlp_bias": false,
|
17 |
"model_type": "llama",
|
18 |
"num_attention_heads": 32,
|
|
|
20 |
"num_key_value_heads": 8,
|
21 |
"pretraining_tp": 1,
|
22 |
"rms_norm_eps": 1e-05,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
"rope_theta": 500000.0,
|
24 |
"tie_word_embeddings": false,
|
25 |
"torch_dtype": "bfloat16",
|