Upload model
Browse files
README.md
CHANGED
|
@@ -342,4 +342,22 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 342 |
### Framework versions
|
| 343 |
|
| 344 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 345 |
- PEFT 0.6.2
|
|
|
|
| 342 |
### Framework versions
|
| 343 |
|
| 344 |
|
| 345 |
+
- PEFT 0.6.2
|
| 346 |
+
## Training procedure
|
| 347 |
+
|
| 348 |
+
|
| 349 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 350 |
+
- load_in_8bit: True
|
| 351 |
+
- load_in_4bit: False
|
| 352 |
+
- llm_int8_threshold: 6.0
|
| 353 |
+
- llm_int8_skip_modules: None
|
| 354 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 355 |
+
- llm_int8_has_fp16_weight: False
|
| 356 |
+
- bnb_4bit_quant_type: fp4
|
| 357 |
+
- bnb_4bit_use_double_quant: False
|
| 358 |
+
- bnb_4bit_compute_dtype: float32
|
| 359 |
+
|
| 360 |
+
### Framework versions
|
| 361 |
+
|
| 362 |
+
|
| 363 |
- PEFT 0.6.2
|