Commit
•
7b757b2
1
Parent(s):
e03974b
Update README.md
Browse files
README.md
CHANGED
@@ -21,18 +21,4 @@ PPL and zero-shot accuracy:
|
|
21 |
The differences between the reported numbers and the reproduced results are possibly variances from the training data processing, seeds, or other random factors.
|
22 |
|
23 |
## Evaluation
|
24 |
-
|
25 |
-
```
|
26 |
-
pip install lm-eval==0.3.0
|
27 |
-
```
|
28 |
-
```
|
29 |
-
python eval_ppl.py --hf_path 1bitLLM/bitnet_b1_58-3B --seqlen 2048
|
30 |
-
```
|
31 |
-
```
|
32 |
-
python eval_task.py --hf_path 1bitLLM/bitnet_b1_58-3B \
|
33 |
-
--batch_size 1 \
|
34 |
-
--tasks \
|
35 |
-
--output_path result.json \
|
36 |
-
--num_fewshot 0 \
|
37 |
-
--ctx_size 2048
|
38 |
-
```
|
|
|
21 |
The differences between the reported numbers and the reproduced results are possibly variances from the training data processing, seeds, or other random factors.
|
22 |
|
23 |
## Evaluation
|
24 |
+
You can now run with model with llama.cpp
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|