Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,18 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
datasets:
|
| 4 |
+
- yuyijiong/Long-Instruction-with-Paraphrasing
|
| 5 |
+
language:
|
| 6 |
+
- zh
|
| 7 |
+
- en
|
| 8 |
+
---
|
| 9 |
+
[Qwen1.5-4B-Chat](https://huggingface.co/Qwen/Qwen1.5-4B-Chat)模型在
|
| 10 |
+
[Long-Instruction-with-Paraphrasing](yuyijiong/Long-Instruction-with-Paraphrasing)数据集上微调，提升了 long-context 能力
|
| 11 |
+
|
| 12 |
+
## Eval on LongBench
|
| 13 |
+
long-context 能力得到提升
|
| 14 |
+
|
| 15 |
+
|model| score| average score|
|
| 16 |
+
|--|--|--|
|
| 17 |
+
|Qwen1.5-4B-Chat|'dureader': 33.61<br>'hotpotqa': 96.5,'lsht': 41.0<br>'multifieldqa_en': 100.0<br>'multifieldqa_zh': 55.4<br>'passage_retrieval_en': 13.0<br>'passage_retrieval_zh': 16.5<br>'qmsum': 22.69<br>'trec': 73.0<br>'vcsum': 15.65|46.73|
|
| 18 |
+
|Qwen1.5-4b-chat-paraph|{'dureader': 31.54<br>'hotpotqa': 99.0<br>'lsht': 39.5<br>'multifieldqa_en': 100.0<br>'multifieldqa_zh': 48.41<br>'passage_retrieval_en': 74.5<br>'passage_retrieval_zh': 62.5<br>'qmsum': 23.9<br>'trec': 74.5<br>'vcsum': 15.79}| **56.96**|
|