enable rope_scaling to allow context lengths up to 32k tokens
Browse files- config.json +4 -1
config.json
CHANGED
|
@@ -18,7 +18,10 @@
|
|
| 18 |
"num_key_value_heads": 8,
|
| 19 |
"pretraining_tp": 1,
|
| 20 |
"rms_norm_eps": 1e-05,
|
| 21 |
-
"rope_scaling":
|
|
|
|
|
|
|
|
|
|
| 22 |
"rope_theta": 500000.0,
|
| 23 |
"tie_word_embeddings": false,
|
| 24 |
"torch_dtype": "float16",
|
|
|
|
| 18 |
"num_key_value_heads": 8,
|
| 19 |
"pretraining_tp": 1,
|
| 20 |
"rms_norm_eps": 1e-05,
|
| 21 |
+
"rope_scaling": {
|
| 22 |
+
"type": "dynamic",
|
| 23 |
+
"factor": 4.0
|
| 24 |
+
},
|
| 25 |
"rope_theta": 500000.0,
|
| 26 |
"tie_word_embeddings": false,
|
| 27 |
"torch_dtype": "float16",
|