Isabelle Cudlitz
commited on
Commit
·
16b2ae0
1
Parent(s):
e87bad0
Upload model
Browse files- README.md +18 -0
- adapter_model.bin +1 -1
README.md
CHANGED
|
@@ -324,4 +324,22 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 324 |
### Framework versions
|
| 325 |
|
| 326 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 327 |
- PEFT 0.6.2
|
|
|
|
| 324 |
### Framework versions
|
| 325 |
|
| 326 |
|
| 327 |
+
- PEFT 0.6.2
|
| 328 |
+
## Training procedure
|
| 329 |
+
|
| 330 |
+
|
| 331 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 332 |
+
- load_in_8bit: False
|
| 333 |
+
- load_in_4bit: True
|
| 334 |
+
- llm_int8_threshold: 6.0
|
| 335 |
+
- llm_int8_skip_modules: None
|
| 336 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 337 |
+
- llm_int8_has_fp16_weight: False
|
| 338 |
+
- bnb_4bit_quant_type: fp4
|
| 339 |
+
- bnb_4bit_use_double_quant: True
|
| 340 |
+
- bnb_4bit_compute_dtype: float32
|
| 341 |
+
|
| 342 |
+
### Framework versions
|
| 343 |
+
|
| 344 |
+
|
| 345 |
- PEFT 0.6.2
|
adapter_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 21020682
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d450184242ed36bc697539c584c275007fc277f73de7d97a292c1b115f020e70
|
| 3 |
size 21020682
|