oldweipro / claude-to-chatgpt Goto Github PK

我自己部署了一套chatgpt-web。如何接入你这个项目呢？

如题，望大佬更新

能不能做个适配？看目前好像是没有这个功能

ERROR: failed to solve: failed to read dockerfile: open /var/lib/docker/tmp/buildkit-mount3997275693/Dockerfile: no such file or directory

Hi, unsure how complicated or even technically possible, but would be appreciated if this could be adapted to accept input from OpenAI-Web GP-4. Hence possibly could work with front-ends that support Uploads via OpenAI-Web calls, eg: GPT-4-Code-Interpreter.
Thankyou for consideration, but completely understand if not possible.

我是用docker方式部署的，原代码无任何修改，dockerfile如下

lllFROM golang:1.20.3-alpine as builder

RUN go env -w GO111MODULE=on
RUN go env -w GOPROXY=https://goproxy.cn,direct

ENV CGO_ENABLED=0

WORKDIR /app

COPY go.mod go.sum ./

RUN go mod tidy && go mod download

COPY . .

RUN go build -ldflags="-w -s" -o /app/claude-to-chatgpt .

FROM alpine:latest

WORKDIR /app

COPY --from=builder /app/claude-to-chatgpt .
COPY --from=builder /app/config.yaml .

EXPOSE 8787

CMD [ "./claude-to-chatgpt", "-c", "config.yaml" ]

0.0.1的时候可以直接部署成功，我也一直在用，后面尝试部署0.0.2后，接口返回出现如下报错

{
  "error": {
    "code": "500",
    "details": "invalid character 'h' looking for beginning of value",
    "message": "Unknown error",
    "param": null,
    "type": "internal_server_error"
  }
}

docker日志报错，请忽略那两个代理报错，我原本不做任何修改想看看还会不会报错json解析问题，代理是可以正常访问claude的：

Unmarshal err: invalid character 'h' looking for beginning of value

目前情况是我即使重新部署0.1版本也是有相同报错，不知道是不是claude的响应结果变了

更新完今天的代码就报这个错了😂

[GIN-debug] Listening and serving HTTP on :8787
Unmarshal err: json: cannot unmarshal object into Go value of type []model.OrganizationsResponse
[GIN] 2023/07/29 - 11:54:12 | 500 | 159.984725ms | 10.10.0.3 | POST "/v1/chat/completions"

如题，应该是调用接口有什么特征泄漏，望大佬更新

详情如图，麻烦您有空的时候帮忙看看~感谢

docker 启动的，latest镜像

有没有可能通过这样的适配，使用Claude2来替代AutoGPT中的ChatGPT4呢？如果可以的话，那项目能够运用的地方就多了。

{"error":{"code":"500","details":"claude错误","message":"claude错误","param":null,"type":"internal_server_error"}}

The cause:

Fix:
See here st1vms/unofficial-claude-api@e2d1b57

好像API无法删除聊天，返回如下错误：
claude-to-chatgpt-api-1 | delete err: delete chat conversations errInternal Server Error xxxxxxxxx
claude-to-chatgpt-api-1 | [GIN] 2023/08/27 - 12:58:17 | 200 | 17.896308744s | 192.168.99.15 | POST "/v1/chat/completions"
claude-to-chatgpt-api-1 | delete err: delete chat conversations errInternal Server Error xxxxx
claude-to-chatgpt-api-1 | [GIN] 2023/08/27 - 12:59:24 | 200 | 5.202486228s | 192.168.99.15 | POST "/v1/chat/completions"

查看网站上有很多标题为“Untitled”的“chat”

访问时候得到如下错误：

测试的py脚本如下

#!/usr/bin/env python3
# -*- encoding: utf-8 -*-
#@time : 2023/08/19 13:23:48

import openai

openai.api_key = ""
openai.api_base = "http://127.0.0.1:8787/v1"

while True:
    text = input("请输入问题：")
    response = openai.ChatCompletion.create(
        model='gpt-4',
        messages=[
            {'role': 'user', 'content': text},
        ],
        stream=True,
        allow_fallback=True
    )

    for chunk in response:
        print(chunk.choices[0].delta.get("content", ""), end="", flush=True)
    print("\n")

如题，使用docker内的ip (https://172.17.0.3:8787/v1) 无法调用，返回错误：

[CHATGPT] Bad Gateway: HTTP code 504 from API

claude2已经推出一段时间了，如题现在项目是否支持claude2

求更新windows打包版，之前的win一键包用不了了

如果我自己写的项目发送的时候带了api key，那我是不是不用在config文件里面去配置api？我现在两端都配置了。如果config不用配置那我就可以去掉了。我自己的项目又key池。

好像claude 对聊天请求数量有限制，聊一段时间就会429，太多的请求。直至等很长时间才恢复，大约等一两小时。

微笑😊
可以支持上上下文，现在都是单聊。
可以支持上调用传递Authorization header
curl --location 'http://api.claude.xxx.com/v1/chat/completions'
--header 'Content-Type: application/json'
--header 'Authorization: Bearer xxx'
--data '{"temperature":0.5,"messages":[{"role":"user","content":"1+1=几"},{"role":"assistant","content":"你确定吗"}],"model":"gpt-3.5-turbo","max_tokens":3000}'