Update max_position_embeddings.

Update the max_position_embeddings to match meta's. It was smaller because this model was finetuned on a smaller context length (8k tokens) than the max (131k).

config.json

@@ -10,7 +10,7 @@
   "hidden_size": 4096,
   "initializer_range": 0.02,
   "intermediate_size": 14336,
-  "max_position_embeddings": 8192,
+  "max_position_embeddings": 131072,
   "mlp_bias": false,
   "model_type": "llama",
   "num_attention_heads": 32,