Upload folder using huggingface_hub

README.md

@@ -25,7 +25,7 @@ base_model:
 ---
 
 # EZCon/SmolVLM2-2.2B-Instruct-4bit-mlx
-This model was converted to MLX format from [`HuggingFaceTB/SmolVLM2-2.2B-Instruct`]() using mlx-vlm version **0.3.2**.
+This model was converted to MLX format from [`HuggingFaceTB/SmolVLM2-2.2B-Instruct`]() using mlx-vlm version **0.3.3**.
 Refer to the [original model card](https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct) for more details on the model.
 ## Use with mlx
 

config.json

@@ -11,6 +11,7 @@
     "decoder_start_token_id": null,
     "diversity_penalty": 0.0,
     "do_sample": false,
+    "dtype": "float32",
     "early_stopping": false,
     "encoder_no_repeat_ngram_size": 0,
     "eos_token_id": null,
@@ -46,11 +47,13 @@
     "pruned_heads": {},
     "quantization": {
         "group_size": 64,
-        "bits": 4
+        "bits": 4,
+        "mode": "affine"
     },
     "quantization_config": {
         "group_size": 64,
-        "bits": 4
+        "bits": 4,
+        "mode": "affine"
     },
     "remove_invalid_values": false,
     "repetition_penalty": 1.0,
@@ -83,7 +86,7 @@
         "return_dict": true,
         "output_hidden_states": false,
         "torchscript": false,
-        "torch_dtype": "bfloat16",
+        "dtype": "bfloat16",
         "pruned_heads": {},
         "tie_word_embeddings": false,
         "chunk_size_feed_forward": 0,
@@ -118,8 +121,6 @@
         "do_sample": false,
         "early_stopping": false,
         "num_beams": 1,
-        "num_beam_groups": 1,
-        "diversity_penalty": 0.0,
         "temperature": 1.0,
         "top_k": 50,
         "top_p": 1.0,
@@ -138,6 +139,8 @@
         "exponential_decay_length_penalty": null,
         "suppress_tokens": null,
         "begin_suppress_tokens": null,
+        "num_beam_groups": 1,
+        "diversity_penalty": 0.0,
         "_name_or_path": "None",
         "_flash_attn_2_enabled": true,
         "model_type": "llama",
@@ -224,27 +227,23 @@
             }
         },
         "use_resampler": false,
-        "tf_legacy_loss": false,
-        "use_bfloat16": false,
         "output_attentions": false
     },
-    "tf_legacy_loss": false,
     "tie_encoder_decoder": false,
     "tie_word_embeddings": false,
     "tokenizer_class": null,
     "top_k": 50,
     "top_p": 1.0,
     "torchscript": false,
-    "transformers_version": "4.56.0.dev0",
+    "transformers_version": "4.57.0.dev0",
     "typical_p": 1.0,
-    "use_bfloat16": false,
     "use_cache": false,
     "use_reentrant_checkpointing": false,
     "vision_config": {
         "return_dict": true,
         "output_hidden_states": false,
         "torchscript": false,
-        "torch_dtype": null,
+        "dtype": null,
         "pruned_heads": {},
         "tie_word_embeddings": false,
         "chunk_size_feed_forward": 0,
@@ -277,8 +276,6 @@
         "do_sample": false,
         "early_stopping": false,
         "num_beams": 1,
-        "num_beam_groups": 1,
-        "diversity_penalty": 0.0,
         "temperature": 1.0,
         "top_k": 50,
         "top_p": 1.0,
@@ -297,6 +294,8 @@
         "exponential_decay_length_penalty": null,
         "suppress_tokens": null,
         "begin_suppress_tokens": null,
+        "num_beam_groups": 1,
+        "diversity_penalty": 0.0,
         "_name_or_path": "",
         "max_image_size": {
             "longest_edge": 384
@@ -306,8 +305,6 @@
             "longest_edge": 1920
         },
         "use_base_siglip": false,
-        "tf_legacy_loss": false,
-        "use_bfloat16": false,
         "hidden_size": 1152,
         "intermediate_size": 4304,
         "num_hidden_layers": 27,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c30695d4dcea2dafa67ee9e3a7bfff906da85d92ae61c22539acc80eef2a53bb
+oid sha256:e407554634bd555aa87388f98b490fd4575463b615c68d7b9e2fb7d30c6add94
 size 1857752703

video_preprocessor_config.json

@@ -28,13 +28,14 @@
     "longest_edge": 384
   },
   "num_frames": 64,
+  "pad_size": null,
   "processor_class": "SmolVLMProcessor",
   "resample": 1,
   "rescale_factor": 0.00392156862745098,
+  "return_metadata": false,
   "size": {
     "longest_edge": 1536
   },
-  "size_divisor": null,
   "video_metadata": null,
   "video_processor_type": "SmolVLMVideoProcessor",
   "video_sampling": {