Upload folder using huggingface_hub

README.md

@@ -5,8 +5,8 @@ tags:
 ---
 
 # moot20/SmolVLM-500M-Base-MLX
-This model was converted to MLX format from [`HuggingFaceTB/SmolVLM-256M-Base`]() using mlx-vlm version **0.1.12**.
-Refer to the [original model card](https://huggingface.co/HuggingFaceTB/SmolVLM-256M-Base) for more details on the model.
+This model was converted to MLX format from [`HuggingFaceTB/SmolVLM-500M-Base`]() using mlx-vlm version **0.1.12**.
+Refer to the [original model card](https://huggingface.co/HuggingFaceTB/SmolVLM-500M-Base) for more details on the model.
 ## Use with mlx
 
 ```bash

config.json

@@ -57,13 +57,13 @@
     "text_config": {
         "vocab_size": 49280,
         "max_position_embeddings": 8192,
-        "hidden_size": 576,
-        "intermediate_size": 1536,
-        "num_hidden_layers": 30,
-        "num_attention_heads": 9,
-        "num_key_value_heads": 3,
+        "hidden_size": 960,
+        "intermediate_size": 2560,
+        "num_hidden_layers": 32,
+        "num_attention_heads": 15,
+        "num_key_value_heads": 5,
         "hidden_act": "silu",
-        "initializer_range": 0.041666666666666664,
+        "initializer_range": 0.02,
         "rms_norm_eps": 1e-05,
         "pretraining_tp": 1,
         "use_cache": true,
@@ -134,7 +134,7 @@
         "decoder_start_token_id": null,
         "task_specific_params": null,
         "problem_type": null,
-        "_name_or_path": "/fsx/m4/experiments/local_experiment_dir/tr_341_vsmollm2_05b/opt_step-25750/unwrapped_model",
+        "_name_or_path": "/fsx/m4/experiments/local_experiment_dir/tr_343_vsmollm2_05b/opt_step-22750/unwrapped_model",
         "_attn_implementation_autoset": false,
         "_flash_attn_2_enabled": true,
         "is_llama_config": true,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cfc4304ed859ecfa88c3b7bc59c9f88bd87a743be95694aea0c7e6c1a1479517
-size 513026363
+oid sha256:3d6caa1cbc55256126a9c584d55cfb1759bb0d359584d1105f111fb48c8a2fac
+size 1015023504

model.safetensors.index.json

@@ -1,6 +1,6 @@
 {
     "metadata": {
-        "total_size": 512969856
+        "total_size": 1014964608
     },
     "weight_map": {
         "connector.modality_projection.proj.weight": "model.safetensors",
@@ -221,6 +221,24 @@
         "language_model.layers.3.self_attn.o_proj.weight": "model.safetensors",
         "language_model.layers.3.self_attn.q_proj.weight": "model.safetensors",
         "language_model.layers.3.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.layers.30.input_layernorm.weight": "model.safetensors",
+        "language_model.layers.30.mlp.down_proj.weight": "model.safetensors",
+        "language_model.layers.30.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.layers.30.mlp.up_proj.weight": "model.safetensors",
+        "language_model.layers.30.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.layers.30.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.layers.30.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.layers.30.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.layers.30.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.layers.31.input_layernorm.weight": "model.safetensors",
+        "language_model.layers.31.mlp.down_proj.weight": "model.safetensors",
+        "language_model.layers.31.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.layers.31.mlp.up_proj.weight": "model.safetensors",
+        "language_model.layers.31.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.layers.31.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.layers.31.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.layers.31.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.layers.31.self_attn.v_proj.weight": "model.safetensors",
         "language_model.layers.4.input_layernorm.weight": "model.safetensors",
         "language_model.layers.4.mlp.down_proj.weight": "model.safetensors",
         "language_model.layers.4.mlp.gate_proj.weight": "model.safetensors",