Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
pipeline_tag: text-generation
|
| 4 |
+
---
|
| 5 |
+
## Training procedure
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 9 |
+
- load_in_8bit: False
|
| 10 |
+
- load_in_4bit: True
|
| 11 |
+
- llm_int8_threshold: 6.0
|
| 12 |
+
- llm_int8_skip_modules: None
|
| 13 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 14 |
+
- llm_int8_has_fp16_weight: False
|
| 15 |
+
- bnb_4bit_quant_type: nf4
|
| 16 |
+
- bnb_4bit_use_double_quant: False
|
| 17 |
+
- bnb_4bit_compute_dtype: float16
|
| 18 |
+
### Framework versions
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
- PEFT 0.4.0.dev0
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
### Notes:
|
| 25 |
+
|
| 26 |
+
fine-tuning parameters:
|
| 27 |
+
|
| 28 |
+
- epochs: 1 (default)
|
| 29 |
+
- learning_rate: 1e-4 (default)
|
| 30 |
+
|
| 31 |
---
|
| 32 |
license: apache-2.0
|
| 33 |
---
|