Commit
•
f5f15b0
1
Parent(s):
3a47346
Update README.md
Browse files
README.md
CHANGED
|
@@ -34,7 +34,22 @@ Llamafactory EVAL
|
|
| 34 |
Humanities: 25.62
|
| 35 |
Other: 27.26
|
| 36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
https://github.com/jzhang38/TinyLlama
|
| 39 |
|
| 40 |
The TinyLlama project aims to **pretrain** a **1.1B Llama model on 3 trillion tokens**. With some proper optimization, we can achieve this within a span of "just" 90 days using 16 A100-40G GPUs 🚀🚀. The training has started on 2023-09-01.
|
|
|
|
| 34 |
Humanities: 25.62
|
| 35 |
Other: 27.26
|
| 36 |
|
| 37 |
+
!CUDA_VISIBLE_DEVICES=0 python src/evaluate.py \
|
| 38 |
+
--model_name_or_path Deathsquad10/TinyLlama-Remix \
|
| 39 |
+
--template vanilla \
|
| 40 |
+
--task cmmlu \
|
| 41 |
+
--split test \
|
| 42 |
+
--lang en \
|
| 43 |
+
--n_shot 5 \
|
| 44 |
+
--use_unsloth \
|
| 45 |
+
--batch_size 2
|
| 46 |
+
|
| 47 |
|
| 48 |
+
Average: 24.98
|
| 49 |
+
STEM: 25.52
|
| 50 |
+
Social Sciences: 24.70
|
| 51 |
+
Humanities: 24.59
|
| 52 |
+
Other: 25.19
|
| 53 |
https://github.com/jzhang38/TinyLlama
|
| 54 |
|
| 55 |
The TinyLlama project aims to **pretrain** a **1.1B Llama model on 3 trillion tokens**. With some proper optimization, we can achieve this within a span of "just" 90 days using 16 A100-40G GPUs 🚀🚀. The training has started on 2023-09-01.
|