AutoGPTQ model for mistralai/Mistral-7B-v0.3: 8bits, gr128, desc_act=False

config.json

@@ -16,6 +16,19 @@
   "num_attention_heads": 32,
   "num_hidden_layers": 32,
   "num_key_value_heads": 8,
+  "quantization_config": {
+    "bits": 8,
+    "checkpoint_format": "gptq",
+    "damp_percent": 0.01,
+    "desc_act": false,
+    "group_size": 128,
+    "model_file_base_name": "gptq_model-8bit-128g",
+    "model_name_or_path": "Mistral-7B-v0.3-gptq-4bit",
+    "quant_method": "gptq",
+    "static_groups": false,
+    "sym": true,
+    "true_sequential": true
+  },
   "rms_norm_eps": 1e-05,
   "rope_theta": 1000000.0,
   "sliding_window": null,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e31960a7b0b7febb8cf860e7f64423438e0bad97d04fffe0a6e44494cd09ffbf
-size 7688169544
+oid sha256:b5309e066356b08e9352f168f77d7cdcb4a7fae624f3af2d2f562085ad3135f4
+size 7685392808

quantize_config.json

@@ -6,6 +6,8 @@
   "static_groups": false,
   "sym": true,
   "true_sequential": true,
-  "model_name_or_path": null,
-  "model_file_base_name": null
+  "model_name_or_path": "Mistral-7B-v0.3-gptq-4bit",
+  "model_file_base_name": "gptq_model-8bit-128g",
+  "quant_method": "gptq",
+  "checkpoint_format": "gptq"
 }