Isabelle Cudlitz
commited on
Commit
·
f5fb524
1
Parent(s):
dc6e80d
Upload model
Browse files- README.md +19 -0
- adapter_model.safetensors +1 -1
README.md
CHANGED
|
@@ -3010,4 +3010,23 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 3010 |
### Framework versions
|
| 3011 |
|
| 3012 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3013 |
- PEFT 0.6.2
|
|
|
|
| 3010 |
### Framework versions
|
| 3011 |
|
| 3012 |
|
| 3013 |
+
- PEFT 0.6.2
|
| 3014 |
+
## Training procedure
|
| 3015 |
+
|
| 3016 |
+
|
| 3017 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 3018 |
+
- quant_method: bitsandbytes
|
| 3019 |
+
- load_in_8bit: True
|
| 3020 |
+
- load_in_4bit: False
|
| 3021 |
+
- llm_int8_threshold: 6.0
|
| 3022 |
+
- llm_int8_skip_modules: None
|
| 3023 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 3024 |
+
- llm_int8_has_fp16_weight: False
|
| 3025 |
+
- bnb_4bit_quant_type: fp4
|
| 3026 |
+
- bnb_4bit_use_double_quant: False
|
| 3027 |
+
- bnb_4bit_compute_dtype: float32
|
| 3028 |
+
|
| 3029 |
+
### Framework versions
|
| 3030 |
+
|
| 3031 |
+
|
| 3032 |
- PEFT 0.6.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 20991608
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fce75774eb0ca99a614392a525d52c901221f1700dad06814833cceca22379f8
|
| 3 |
size 20991608
|