Update configuration_llama.py, required to get model to Load as `rope_scaling` needs to be None, or else a dictionary (#1)

- Update configuration_llama.py, required to get model to Load as `rope_scaling` needs to be None, or else a dictionary (4d436dbbce1605d4eb4f290648f02446eaac372f)


Co-authored-by: Tom Jobbins <[email protected]>

configuration_llama.py

@@ -124,7 +124,7 @@ class LlamaConfig(PretrainedConfig):
         pretraining_tp=1,
         tie_word_embeddings=False,
         rope_theta=10000,
-        rope_scaling="yarn",
+        rope_scaling=None,
         attention_bias=False,
         **kwargs,
     ):