princeton-nlp
/

FullAttention-Llama-2-7b-6k

Inference Endpoints

text-generation-inference

Model card Files Files and versions Community

princeton-nlp commited on Oct 26, 2023

Commit

2ff0078

•

1 Parent(s): a4d455c

Update config.json

Updating RoPE theta value.

Files changed (1) hide show

config.json +1 -1

config.json CHANGED Viewed

@@ -18,7 +18,7 @@
   "pretraining_tp": 1,
   "rms_norm_eps": 1e-05,
   "rope_scaling": null,
-  "rope_theta": 10000.0,
   "summary_length": 0,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",

   "pretraining_tp": 1,
   "rms_norm_eps": 1e-05,
   "rope_scaling": null,
+  "rope_theta": 80000.0,
   "summary_length": 0,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",