Update config.json
Browse files- config.json +6 -2
config.json
CHANGED
@@ -10,7 +10,7 @@
|
|
10 |
"hidden_size": 4096,
|
11 |
"initializer_range": 0.02,
|
12 |
"intermediate_size": 12288,
|
13 |
-
"max_position_embeddings":
|
14 |
"max_window_layers": 36,
|
15 |
"model_type": "qwen3",
|
16 |
"num_attention_heads": 32,
|
@@ -18,7 +18,11 @@
|
|
18 |
"num_key_value_heads": 8,
|
19 |
"pad_token_id": 151654,
|
20 |
"rms_norm_eps": 1e-06,
|
21 |
-
"rope_scaling":
|
|
|
|
|
|
|
|
|
22 |
"rope_theta": 1000000,
|
23 |
"sliding_window": null,
|
24 |
"tie_word_embeddings": false,
|
|
|
10 |
"hidden_size": 4096,
|
11 |
"initializer_range": 0.02,
|
12 |
"intermediate_size": 12288,
|
13 |
+
"max_position_embeddings": 81920,
|
14 |
"max_window_layers": 36,
|
15 |
"model_type": "qwen3",
|
16 |
"num_attention_heads": 32,
|
|
|
18 |
"num_key_value_heads": 8,
|
19 |
"pad_token_id": 151654,
|
20 |
"rms_norm_eps": 1e-06,
|
21 |
+
"rope_scaling": {
|
22 |
+
"rope_type": "yarn",
|
23 |
+
"factor": 2.0,
|
24 |
+
"original_max_position_embeddings": 41960
|
25 |
+
},
|
26 |
"rope_theta": 1000000,
|
27 |
"sliding_window": null,
|
28 |
"tie_word_embeddings": false,
|