这时FastGPT只会提示“LLM model response empty” OneAPI返回的明细如下 // event1{"choices": [ {"content_filter_results": {},"delta": {"content":"","role":"assistant"},"finish_reason":null,"index":0,"logprobs":null} ],"created":1724116111,"id":"chatcmpl-9y7eJyXhFvB05DLc6oltsQw...
MODEL_PATH)# set Embedding Model pathEMBEDDING_PATH=os.environ.get('EMBEDDING_PATH','BAAI/bge-m3')@asynccontextmanagerasyncdeflifespan(app:FastAPI):yieldiftorch.cuda.is_available():torch.cuda.empty_cache()torch.cuda.ipc_collect()app=FastAPI(lifespan=life...
{ openapiPrefix: 'fastgpt', vectorMaxProcess: 15, qaMaxProcess: 15, pgHNSWEfSearch: 100 }, subPlans: undefined, llmModels: [ { model: 'Qwen1.5-14B-Chat', name: '通义千问1.5-14B-Chat', maxContext: 32768, avatar: '/imgs/model/openai.svg', maxResponse: 8000, quoteMaxToken: 20000...
"llmModels": [ { "model": "gpt-3.5-turbo", "name": "gpt-3.5-turbo", "maxContext": 16000, "avatar": "/imgs/model/openai.svg", "maxResponse": 4000, "quoteMaxToken": 13000, "maxTemperature": 1.2, "charsPointsPrice": 0, "censor": false, "vision": false, "datasetProcess": tru...
2、(可选)启动 LLM 聊天服务 text-generation-webui ### 2-1、(可选)启动 LLM 聊天服务 【fastgpt】+【one-api】+【Xinference】fastgpt:https://github.com/labring/FastGPT one-api:https://github.com/songquanpeng/one-api Xinference:
我也是报错llm输出是empty,且有时候会输出Action内容,内容截断的有问题。但都不是必显,有几率出现。我是qwen1.5-72b,xinference=0.13.0,fastgpt=4.8.9. Sign up for freeto join this conversation on GitHub. Already have an account?Sign in to comment...
jsonRes<InitDateResponse>(res, { data: { @@ -37,148 +26,3 @@ async function handler(req: NextApiRequest, res: NextApiResponse) { } export default handler; const defaultFeConfigs: FastGPTFeConfigsType = { show_emptyChat: true, show_git: true, docUrl: 'https://doc.fastgpt.in', ...
maxTokens: toolModel.maxContext - 500 // filter token. not response maxToken }); const requestMessages = await loadRequestMessages(filterMessages); // console.log(JSON.stringify(filterMessages, null, 2)); /* Run llm */ const ai = getAIApi({ @@ -98,7 +100,7 @@ export const runToo...
"default_response": "Default Response", "edit_api_key": "Edit Key information", "edit_link": "Edit", "feishu_name": "Lark", "link_name": "Name of the share link", "qpm_tips": "How many times per minute can each IP ask at most", "qpm_is_empty": "QPM cannot be empty", ...
Automatic Object Completion: Ask LLM to fill empty fields of object Dynamic Object Modification: Ask LLM to expand or modify property of object Sentiment Analysis and Property Assignment: Ask LLM to check sentiments (positive/negitive) on object and fill the property. For example, sentiment ...