root commited on
Commit ·
d3ef0a4
1
Parent(s): 56ce0a4
fix base config.json rope scale 2
Browse files- config.json +4 -4
config.json
CHANGED
|
@@ -6,17 +6,17 @@
|
|
| 6 |
"hidden_size": 2304,
|
| 7 |
"initializer_range": 0.02,
|
| 8 |
"intermediate_size": 6144,
|
| 9 |
-
"max_position_embeddings":
|
| 10 |
"num_attention_heads": 18,
|
| 11 |
"num_hidden_layers": 28,
|
| 12 |
"num_key_value_heads": 18,
|
| 13 |
"pad_token_id": 0,
|
| 14 |
"pretraining_tp": 1,
|
| 15 |
"rms_norm_eps": 1e-06,
|
|
|
|
| 16 |
"tie_word_embeddings": false,
|
| 17 |
"transformers_version": "4.31.0",
|
| 18 |
"use_cache": true,
|
| 19 |
-
"
|
| 20 |
-
"
|
| 21 |
}
|
| 22 |
-
|
|
|
|
| 6 |
"hidden_size": 2304,
|
| 7 |
"initializer_range": 0.02,
|
| 8 |
"intermediate_size": 6144,
|
| 9 |
+
"max_position_embeddings": 8192,
|
| 10 |
"num_attention_heads": 18,
|
| 11 |
"num_hidden_layers": 28,
|
| 12 |
"num_key_value_heads": 18,
|
| 13 |
"pad_token_id": 0,
|
| 14 |
"pretraining_tp": 1,
|
| 15 |
"rms_norm_eps": 1e-06,
|
| 16 |
+
"rope_scaling": {"type":"linear", "factor": 2.0 },
|
| 17 |
"tie_word_embeddings": false,
|
| 18 |
"transformers_version": "4.31.0",
|
| 19 |
"use_cache": true,
|
| 20 |
+
"torch_dtype": "bfloat16",
|
| 21 |
+
"vocab_size": 96000
|
| 22 |
}
|
|
|