Model save
Browse files
README.md
CHANGED
|
@@ -107,6 +107,17 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 107 |
- bnb_4bit_use_double_quant: False
|
| 108 |
- bnb_4bit_compute_dtype: float16
|
| 109 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 110 |
The following `bitsandbytes` quantization config was used during training:
|
| 111 |
- load_in_8bit: False
|
| 112 |
- load_in_4bit: True
|
|
@@ -143,6 +154,7 @@ The following hyperparameters were used during training:
|
|
| 143 |
- PEFT 0.4.0
|
| 144 |
- PEFT 0.4.0
|
| 145 |
- PEFT 0.4.0
|
|
|
|
| 146 |
- Transformers 4.31.0
|
| 147 |
- Pytorch 2.0.1+cu118
|
| 148 |
- Datasets 2.14.5
|
|
|
|
| 107 |
- bnb_4bit_use_double_quant: False
|
| 108 |
- bnb_4bit_compute_dtype: float16
|
| 109 |
|
| 110 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 111 |
+
- load_in_8bit: False
|
| 112 |
+
- load_in_4bit: True
|
| 113 |
+
- llm_int8_threshold: 6.0
|
| 114 |
+
- llm_int8_skip_modules: None
|
| 115 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 116 |
+
- llm_int8_has_fp16_weight: False
|
| 117 |
+
- bnb_4bit_quant_type: nf4
|
| 118 |
+
- bnb_4bit_use_double_quant: False
|
| 119 |
+
- bnb_4bit_compute_dtype: float16
|
| 120 |
+
|
| 121 |
The following `bitsandbytes` quantization config was used during training:
|
| 122 |
- load_in_8bit: False
|
| 123 |
- load_in_4bit: True
|
|
|
|
| 154 |
- PEFT 0.4.0
|
| 155 |
- PEFT 0.4.0
|
| 156 |
- PEFT 0.4.0
|
| 157 |
+
- PEFT 0.4.0
|
| 158 |
- Transformers 4.31.0
|
| 159 |
- Pytorch 2.0.1+cu118
|
| 160 |
- Datasets 2.14.5
|