Upload folder using huggingface_hub

chat_template.jinja

@@ -293,4 +293,4 @@
     {%- endif %}
 {%- endfor %}
 {#- Copyright 2025-present the Unsloth team. All rights reserved. #}
-{#- Licensed under the Apache License, Version 2.0 (the "License") #}
+{#- Licensed under the Apache License, Version 2.0 (the "License") #}

config.json

@@ -202,6 +202,9 @@
       "model.vision_tower.transformer.layers.23.attention.v_proj",
       "model.vision_tower.transformer.layers.23.attention.q_proj",
       "model.vision_tower.transformer.layers.23.attention.o_proj",
+      "model.multi_modal_projector.patch_merger.merging_layer",
+      "model.multi_modal_projector.linear_1",
+      "model.multi_modal_projector.linear_2",
       "lm_head"
     ],
     "kv_cache_scheme": null,

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0849ca25671278e72c5cac219b85eec06405624859f306cd8907756f724aa9bc
-size 4946472576
+oid sha256:60ad56ab73386cdda793c4e0ca7f5b238b0bd0e8c85cefa5a52b925f5ff63440
+size 4999392136

model.safetensors.index.json

@@ -1,6 +1,6 @@
 {
   "metadata": {
-    "total_size": 14972430768
+    "total_size": 15025351040
   },
   "weight_map": {
     "language_model.lm_head.weight": "model-00004-of-00004.safetensors",
@@ -926,16 +926,10 @@
     "language_model.model.layers.9.self_attn.v_proj.weight_scale": "model-00001-of-00004.safetensors",
     "language_model.model.layers.9.self_attn.v_proj.weight_shape": "model-00001-of-00004.safetensors",
     "language_model.model.norm.weight": "model-00003-of-00004.safetensors",
-    "multi_modal_projector.linear_1.weight_packed": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.linear_1.weight_scale": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.linear_1.weight_shape": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.linear_2.weight_packed": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.linear_2.weight_scale": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.linear_2.weight_shape": "model-00001-of-00004.safetensors",
+    "multi_modal_projector.linear_1.weight": "model-00001-of-00004.safetensors",
+    "multi_modal_projector.linear_2.weight": "model-00001-of-00004.safetensors",
     "multi_modal_projector.norm.weight": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.patch_merger.merging_layer.weight_packed": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.patch_merger.merging_layer.weight_scale": "model-00001-of-00004.safetensors",
-    "multi_modal_projector.patch_merger.merging_layer.weight_shape": "model-00001-of-00004.safetensors",
+    "multi_modal_projector.patch_merger.merging_layer.weight": "model-00001-of-00004.safetensors",
     "vision_tower.ln_pre.weight": "model-00001-of-00004.safetensors",
     "vision_tower.patch_conv.weight": "model-00001-of-00004.safetensors",
     "vision_tower.transformer.layers.0.attention.k_proj.weight": "model-00001-of-00004.safetensors",

recipe.yaml

@@ -2,7 +2,7 @@ default_stage:
   default_modifiers:
     AWQModifier:
       targets: [Linear]
-      ignore: [lm_head, 're:.*vision_tower.*', 're:.*multi_modal_projection.*']
+      ignore: [lm_head, 're:.*vision_tower.*', 're:.*multi_modal_projector.*']
       scheme: W4A16
       mappings:
       - smooth_layer: re:.*input_layernorm$