Upload model
Browse files
README.md
CHANGED
|
@@ -16,6 +16,18 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 16 |
- bnb_4bit_use_double_quant: False
|
| 17 |
- bnb_4bit_compute_dtype: float32
|
| 18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
The following `bitsandbytes` quantization config was used during training:
|
| 20 |
- quant_method: bitsandbytes
|
| 21 |
- load_in_8bit: False
|
|
@@ -29,6 +41,7 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 29 |
- bnb_4bit_compute_dtype: float16
|
| 30 |
### Framework versions
|
| 31 |
|
|
|
|
| 32 |
- PEFT 0.6.0.dev0
|
| 33 |
|
| 34 |
- PEFT 0.4.0
|
|
|
|
| 16 |
- bnb_4bit_use_double_quant: False
|
| 17 |
- bnb_4bit_compute_dtype: float32
|
| 18 |
|
| 19 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 20 |
+
- quant_method: bitsandbytes
|
| 21 |
+
- load_in_8bit: True
|
| 22 |
+
- load_in_4bit: False
|
| 23 |
+
- llm_int8_threshold: 6.0
|
| 24 |
+
- llm_int8_skip_modules: None
|
| 25 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 26 |
+
- llm_int8_has_fp16_weight: False
|
| 27 |
+
- bnb_4bit_quant_type: fp4
|
| 28 |
+
- bnb_4bit_use_double_quant: False
|
| 29 |
+
- bnb_4bit_compute_dtype: float32
|
| 30 |
+
|
| 31 |
The following `bitsandbytes` quantization config was used during training:
|
| 32 |
- quant_method: bitsandbytes
|
| 33 |
- load_in_8bit: False
|
|
|
|
| 41 |
- bnb_4bit_compute_dtype: float16
|
| 42 |
### Framework versions
|
| 43 |
|
| 44 |
+
- PEFT 0.6.0.dev0
|
| 45 |
- PEFT 0.6.0.dev0
|
| 46 |
|
| 47 |
- PEFT 0.4.0
|