Update config.json

config.json

@@ -1,9 +1,14 @@
 {
-    "_commit_hash": null,
-    "_name_or_path": "../pixtral",
+    "_name_or_path": "mgoin/pixtral-12b",
     "architectures": [
         "LlavaForConditionalGeneration"
     ],
+    "ignore_index": -100,
+    "image_seq_length": 1,
+    "image_token_index": 10,
+    "model_type": "llava",
+    "projector_hidden_act": "gelu",
+    "initializer_range": 0.02,
     "quantization_config": {
         "config_groups": {
             "group_0": {
@@ -227,35 +232,35 @@
         "quant_method": "compressed-tensors",
         "quantization_status": "compressed"
     },
-    "ignore_index": -100,
-    "image_seq_length": 1,
-    "image_token_index": 10,
-    "model_type": "llava",
-    "projector_hidden_act": "gelu",
     "text_config": {
+        "_attn_implementation_autoset": true,
         "hidden_size": 5120,
-        "head_dim": 128,
         "intermediate_size": 14336,
-        "is_composition": true,
+        "head_dim": 128,
         "max_position_embeddings": 1024000,
+        "is_composition": true,
         "model_type": "mistral",
         "num_hidden_layers": 40,
+        "num_attention_heads": 32,
         "num_key_value_heads": 8,
-        "rms_norm_eps": 0.00001,
-        "rope_theta": 1000000000,
+        "rms_norm_eps": 1e-05,
+        "rope_theta": 1000000000.0,
         "sliding_window": null,
         "vocab_size": 131072
     },
     "torch_dtype": "bfloat16",
-    "transformers_version": null,
+    "transformers_version": "4.47.1",
     "vision_config": {
         "head_dim": 64,
         "hidden_act": "silu",
         "image_size": 1024,
+        "intermediate_size": 4096,
         "is_composition": true,
         "model_type": "pixtral",
         "patch_size": 16,
-        "rope_theta": 10000,
+        "num_attention_heads": 16,
+        "num_hidden_layers": 24,
+        "rope_theta": 10000.0,
         "tie_word_embeddings": false
     },
     "vision_feature_layer": -1,