use NousResearch/Meta-Llama-3.1-70B-Instruct as tinygrad llama-3.1-70b model, previously using non-instruct model

exo/models.py

@@ -8,7 +8,7 @@ model_base_shards = {
   },
   "llama-3.1-70b": {
     "MLXDynamicShardInferenceEngine": Shard(model_id="mlx-community/Meta-Llama-3.1-70B-Instruct-4bit", start_layer=0, end_layer=0, n_layers=80),
-    "TinygradDynamicShardInferenceEngine": Shard(model_id="NousResearch/Meta-Llama-3.1-70B", start_layer=0, end_layer=0, n_layers=80),
+    "TinygradDynamicShardInferenceEngine": Shard(model_id="NousResearch/Meta-Llama-3.1-70B-Instruct", start_layer=0, end_layer=0, n_layers=80),
   },
   "llama-3.1-405b": {"MLXDynamicShardInferenceEngine": Shard(model_id="mlx-community/Meta-Llama-3.1-405B-4bit", start_layer=0, end_layer=0, n_layers=126),},
   "llama-3-8b": {