Upload folder using huggingface_hub (#4)

- 357341fe29bdeaf6de6a364238dbb4523fa567a31cb86c4d861b602014161642 (22a38264bf9f20170b88e67a5dfb59c7c6d0d0e9)

config.json

@@ -1,6 +1,6 @@
 {
   "_attn_implementation_autoset": true,
-  "_name_or_path": "HuggingFaceTB/SmolLM2-1.7B-Instruct",
+  "_name_or_path": "/tmp/models/tmpj8yy9slv/tmpg5m0u7si",
   "architectures": [
     "LlamaForCausalLM"
   ],
@@ -21,6 +21,26 @@
   "num_key_value_heads": 32,
   "pad_token_id": 2,
   "pretraining_tp": 1,
+  "quantization_config": {
+    "quant_config": {
+      "offload_meta": false,
+      "scale_quant_params": null,
+      "weight_quant_params": {
+        "axis": 1,
+        "channel_wise": true,
+        "group_size": 64,
+        "nbits": 4,
+        "optimize": true,
+        "round_zero": true,
+        "view_as_float": false
+      },
+      "zero_quant_params": null
+    },
+    "quant_method": "hqq",
+    "skip_modules": [
+      "lm_head"
+    ]
+  },
   "rms_norm_eps": 1e-05,
   "rope_scaling": null,
   "rope_theta": 130000,

qmodel.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e73582d98e0303a1e77477ad769dd6f2ef63610a3932cba84f84a9e51a24cdbe
-size 1107676912
+oid sha256:11c96faf0e2f45fa9d776bd647260422f2045033d2e00bc1ba212f9bcb8c8a6c
+size 1409859568

smash_config.json

@@ -11,7 +11,7 @@
     "quant_hqq_weight_bits": 4,
     "max_batch_size": 1,
     "device": "cuda",
-    "cache_dir": "/tmp/models/tmpimi9ply8",
+    "cache_dir": "/tmp/models/tmpj8yy9slv",
     "task": "",
     "save_load_fn": "hqq",
     "save_load_fn_args": {},