无法索引知识库 - Githubissues

weirdorr commented 3 months ago

例行检查

[ ] 我已确认目前没有类似 issue
[ ] 我已完整查看过项目 README，以及项目文档
[ ] 我使用了自己的 key，并确认我的 key 是可正常使用的
[ ] 我理解并愿意跟进此 issue，协助测试和提供反馈
[x] 我理解并认可上述内容，并理解项目维护者精力有限，不遵循规则的 issue 可能会被无视或直接关闭

你的版本

[ ] 公有云版本
[x] 私有部署版本, 具体版本号:

问题描述, 日志截图

复现步骤

预期结果

相关截图 error => 400 Inference error: Range of input length should be [1, 2048] (request id: 2024070208305054674516722308854) [Error] 2024-07-02 08:30:50 400 Inference error: Range of input length should be [1, 2048] (request id: 2024070208305054674516722308854) [Info] 2024-07-02 08:30:50 [Vector Queue] Start [Info] 2024-07-02 08:30:50 [Vector Queue] Start Embedding Error eF [Error]: 400 Inference error: Range of input length should be [1, 2048] (request id: 2024070208305063434452547591163) at eL.generate (/app/projects/app/.next/server/chunks/76750.js:15:67594) at av.makeStatusError (/app/projects/app/.next/server/chunks/76750.js:15:79337) at av.makeRequest (/app/projects/app/.next/server/chunks/76750.js:15:80260) at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at async c (/app/projects/app/.next/server/chunks/75612.js:166:1395) at async f (/app/projects/app/.next/server/chunks/75612.js:137:15493) at async Promise.all (index 0) at async u (/app/projects/app/.next/server/chunks/75612.js:314:30962) at async /app/projects/app/.next/server/chunks/75612.js:316:3257 at async o (/app/projects/app/.next/server/chunks/75612.js:134:21731) { status: 400, headers: { 'content-length': '200', 'content-type': 'application/json; charset=utf-8', date: 'Tue, 02 Jul 2024 08:30:50 GMT', 'x-oneapi-request-id': '2024070208305063434452547591163' }, error: { message: 'Inference error: Range of input length should be [1, 2048] (request id: 2024070208305063434452547591163)', type: 'upstream_error', param: '400', code: 'bad_response_status_code' }, code: 'bad_response_status_code', param: '400', type: 'upstream_error' }

c121914yu commented 3 months ago

模型维度大于 1536 ，目前系统只能使用<=1536 维度的向量模型

leavegee commented 2 months ago

1536也报错呀~ 阿里的text-embedding-v2

raozhiyong11 commented 1 month ago

@leavegee @c121914yu @weirdorr 请问这个问题怎么解决的？我也遇到一样的问题，而且阿里云后台一直在扣费，是不是索引不成功会一直尝试？

weirdorr commented 1 month ago

@raozhiyong11 试试改一下max-token的值

raozhiyong11 commented 1 month ago

@raozhiyong11 试试改一下max-token的值

已经改过了，之前默认是3000，现在改成2000了重新启动还是不行，关键是一直重试调用阿里的text-embedding模型

labring / FastGPT

无法索引知识库 #1912