Add W4A16 AutoRound: quantization_config.json
Browse files
W4A16-AutoRound/quantization_config.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bits": 4,
|
| 3 |
+
"group_size": 128,
|
| 4 |
+
"sym": true,
|
| 5 |
+
"data_type": "int",
|
| 6 |
+
"iters": 500,
|
| 7 |
+
"nsamples": 512,
|
| 8 |
+
"autoround_version": "0.9.5",
|
| 9 |
+
"quant_method": "auto-round",
|
| 10 |
+
"packing_format": "auto_round:auto_gptq"
|
| 11 |
+
}
|