Upload quant_config.txt with huggingface_hub
Browse files- quant_config.txt +15 -0
quant_config.txt
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
BitsAndBytesConfig {
|
| 2 |
+
"_load_in_4bit": true,
|
| 3 |
+
"_load_in_8bit": false,
|
| 4 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 5 |
+
"bnb_4bit_quant_storage": "uint8",
|
| 6 |
+
"bnb_4bit_quant_type": "fp4",
|
| 7 |
+
"bnb_4bit_use_double_quant": false,
|
| 8 |
+
"llm_int8_enable_fp32_cpu_offload": false,
|
| 9 |
+
"llm_int8_has_fp16_weight": false,
|
| 10 |
+
"llm_int8_skip_modules": null,
|
| 11 |
+
"llm_int8_threshold": 6.0,
|
| 12 |
+
"load_in_4bit": true,
|
| 13 |
+
"load_in_8bit": false,
|
| 14 |
+
"quant_method": "bitsandbytes"
|
| 15 |
+
}
|