Update README.md
Browse files
README.md
CHANGED
|
@@ -3,4 +3,36 @@ language:
|
|
| 3 |
- en
|
| 4 |
datasets:
|
| 5 |
- Glavin001/generate-questions-v0.1
|
| 6 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
- en
|
| 4 |
datasets:
|
| 5 |
- Glavin001/generate-questions-v0.1
|
| 6 |
+
library_name: peft
|
| 7 |
+
---
|
| 8 |
+
## Training procedure
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 12 |
+
- quant_method: bitsandbytes
|
| 13 |
+
- load_in_8bit: True
|
| 14 |
+
- load_in_4bit: False
|
| 15 |
+
- llm_int8_threshold: 6.0
|
| 16 |
+
- llm_int8_skip_modules: None
|
| 17 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 18 |
+
- llm_int8_has_fp16_weight: False
|
| 19 |
+
- bnb_4bit_quant_type: fp4
|
| 20 |
+
- bnb_4bit_use_double_quant: False
|
| 21 |
+
- bnb_4bit_compute_dtype: float32
|
| 22 |
+
|
| 23 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 24 |
+
- quant_method: bitsandbytes
|
| 25 |
+
- load_in_8bit: True
|
| 26 |
+
- load_in_4bit: False
|
| 27 |
+
- llm_int8_threshold: 6.0
|
| 28 |
+
- llm_int8_skip_modules: None
|
| 29 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 30 |
+
- llm_int8_has_fp16_weight: False
|
| 31 |
+
- bnb_4bit_quant_type: fp4
|
| 32 |
+
- bnb_4bit_use_double_quant: False
|
| 33 |
+
- bnb_4bit_compute_dtype: float32
|
| 34 |
+
### Framework versions
|
| 35 |
+
|
| 36 |
+
- PEFT 0.6.0.dev0
|
| 37 |
+
|
| 38 |
+
- PEFT 0.6.0.dev0
|