Commit
•
6dc8bae
1
Parent(s):
3199f2a
Upload model
Browse files
README.md
CHANGED
|
@@ -15,6 +15,17 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 15 |
- bnb_4bit_use_double_quant: False
|
| 16 |
- bnb_4bit_compute_dtype: float16
|
| 17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
The following `bitsandbytes` quantization config was used during training:
|
| 19 |
- load_in_8bit: False
|
| 20 |
- load_in_4bit: True
|
|
@@ -27,6 +38,7 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 27 |
- bnb_4bit_compute_dtype: float16
|
| 28 |
### Framework versions
|
| 29 |
|
|
|
|
| 30 |
- PEFT 0.5.0.dev0
|
| 31 |
|
| 32 |
- PEFT 0.5.0.dev0
|
|
|
|
| 15 |
- bnb_4bit_use_double_quant: False
|
| 16 |
- bnb_4bit_compute_dtype: float16
|
| 17 |
|
| 18 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 19 |
+
- load_in_8bit: False
|
| 20 |
+
- load_in_4bit: True
|
| 21 |
+
- llm_int8_threshold: 6.0
|
| 22 |
+
- llm_int8_skip_modules: None
|
| 23 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 24 |
+
- llm_int8_has_fp16_weight: False
|
| 25 |
+
- bnb_4bit_quant_type: nf4
|
| 26 |
+
- bnb_4bit_use_double_quant: False
|
| 27 |
+
- bnb_4bit_compute_dtype: float16
|
| 28 |
+
|
| 29 |
The following `bitsandbytes` quantization config was used during training:
|
| 30 |
- load_in_8bit: False
|
| 31 |
- load_in_4bit: True
|
|
|
|
| 38 |
- bnb_4bit_compute_dtype: float16
|
| 39 |
### Framework versions
|
| 40 |
|
| 41 |
+
- PEFT 0.5.0.dev0
|
| 42 |
- PEFT 0.5.0.dev0
|
| 43 |
|
| 44 |
- PEFT 0.5.0.dev0
|