Initial upload of Floppa-12B (GPTQ)

config.json

@@ -44,7 +44,9 @@
     },
     "format": "pack-quantized",
     "global_compression_ratio": null,
-    "ignore": [],
+    "ignore": [
+      "embed_out"
+    ],
     "kv_cache_scheme": null,
     "quant_method": "compressed-tensors",
     "quantization_status": "compressed",

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a92b7acbca09d0002deb4e3c8d8718e71c9c963505d3c384c0e9d3a414e576a6
+oid sha256:3bd218eb9562e54366f50146a568dc75e3a20f14cbc219e5d8155de77272d33c
 size 4999615544

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d81bba0d0ba6568f027673c5c78ab7d6322cc2817584e97da33f7acf0f0d6010
-size 2202264608
+oid sha256:a86448798309d43afa484d8bedfc3a81f857b6b4e6105c2c3126bc9870412171
+size 3102483856

model.safetensors.index.json

@@ -1,12 +1,10 @@
 {
   "metadata": {
-    "total_parameters": 1800448290,
-    "total_size": 7201794320
+    "total_parameters": 2025503008,
+    "total_size": 8102013184
   },
   "weight_map": {
-    "embed_out.weight_packed": "model-00002-of-00002.safetensors",
-    "embed_out.weight_scale": "model-00002-of-00002.safetensors",
-    "embed_out.weight_shape": "model-00002-of-00002.safetensors",
+    "embed_out.weight": "model-00002-of-00002.safetensors",
     "gpt_neox.embed_in.weight": "model-00001-of-00002.safetensors",
     "gpt_neox.final_layer_norm.bias": "model-00002-of-00002.safetensors",
     "gpt_neox.final_layer_norm.weight": "model-00002-of-00002.safetensors",

recipe.yaml

@@ -2,7 +2,7 @@ default_stage:
   default_modifiers:
     GPTQModifier:
       targets: [Linear]
-      ignore: [lm_head]
+      ignore: [lm_head, embed_out]
       scheme: W4A16
       block_size: 128
       dampening_frac: 0.01