Upload MistralForCausalLM

config.json

@@ -4,7 +4,7 @@
   "activation_dropout": 0.0,
   "activation_function": "relu",
   "architectures": [
-    "OPTForCausalLM"
+    "MistralForCausalLM"
   ],
   "attention_dropout": 0.0,
   "bos_token_id": 2,
@@ -13,16 +13,24 @@
   "enable_bias": true,
   "eos_token_id": 2,
   "ffn_dim": 3072,
+  "hidden_act": "silu",
   "hidden_size": 768,
   "init_std": 0.02,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
   "layer_norm_elementwise_affine": true,
   "layerdrop": 0.0,
   "max_position_embeddings": 2048,
-  "model_type": "opt",
+  "model_type": "mistral",
   "num_attention_heads": 12,
   "num_hidden_layers": 12,
+  "num_key_value_heads": 8,
   "pad_token_id": 1,
   "prefix": "</s>",
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "tie_word_embeddings": false,
   "torch_dtype": "float32",
   "transformers_version": "4.34.0",
   "use_cache": true,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d49f3e861a560dd71a0cc1a885d918a3b709033f1244402b8e394155d39f7c9b
-size 501021082
+oid sha256:cd2429156f122a87307530b1222cb15b7726440278647f72d1e8833784686b34
+size 1988802630