-
With the release of the new SWE-bench evaluation harness last month, we have recently put forth a new set of submission guidelines requirements, detailed fully in the README and [here](https://www.swe…
-
I think a simple solution could be to wait until the last chunk arrived and then calculate the usage with the prompt and response strings as shown here: https://github.com/openai/openai-cookbook/blob/…
-
-
Experiment 1
| Prompt A | GPT-3.5-Turbo-0613 (Temperature=1) |
|----------|----------------------------------|
| 1 | Token wise but penny foolish. |
| 2 | Token wise but penny fooli…
-
hello,
我想咨询一下在MT-bench上测试时,使用的reference answer 是通过 gen_api_answer.py --model gpt-4-0125-preview这个命令来获取的吗?
生成的reference answer有80个,然后把其中100~130个用official comment[https://github.com/lm-sys/FastChat/…
-
主题 berry 启用方式:启动的时候添加环境变量 THEME=berry 即可,未来会允许直接在后台改
感谢主题作者:https://github.com/MartialBE
欢迎在此提交相关问题。
---
2024年4月5日:berry 主题目前相比默认主题缺失的功能:
1. 令牌维度模型限制;
2. 令牌维度 ip 限制;
3. 渠道批量添加 key;
4. …
-
I noticed that in the recent commit (dc4d1c6b8d0d52ae5897e8203912b9b047ac981f) on https://github.com/alexrudall/ruby-openai, the processing for JSON.parse within the json_post method seems to have bee…
-
`2023-08-05 21:42:12,422 - INFO - Updata config: {'HyDE': False, 'answer_depth': 1.0, 'ask_topk': 5, 'bases_root': 'bases', 'enable_ask': True, 'enable_config': True, 'enable_review': True, 'enable_se…
-
### Confirm this is a Node library issue and not an underlying OpenAI API issue
- [X] This is an issue with the Node library
### Describe the bug
This is an example from the [api-reference](https:/…
-
### Bug Details
Currently am trying to deploy a new instance of the PubSec-Info-Assist and the enrichment web app doesn't come online, and deployment gets stuck in an infinite loop.
**Steps To R…