indexing error

[gerrampard]

indexing error: type error: fetch failed after initialing vector database, (tried several models, MacOS 14.3.1, Reor 0.1.45)

[kankanai]

me too

[imb4dm4n]

+1

[samlhuillier]

I've made several changes to to indexing in the new version: https://github.com/reorproject/reor/releases/tag/v0.1.46 Could you guys try it and let me know if you still get this error

[gerrampard]

I've made several changes to to indexing in the new version: https://github.com/reorproject/reor/releases/tag/v0.1.46 Could you guys try it and let me know if you still get this error

unfortunately, still same error, and now app crash! (submitted the crash report to apple, maybe you can take a look)

[samlhuillier]

Would you be able to run from source and send me what shows up in logs, just I don't have much to debug with right now. If you install the latest version of node and run:

git clone https://github.com/reorproject/reor
npm install
npm run dev

that would be much appreciated :)

[Shiye-10Pages]

@samlhuillier

[samlhuillier]

@SYLJX you need to install node js: https://nodejs.org/en/download

[kankanai]

Would you be able to run from source and send me what shows up in logs, just I don't have much to debug with right now. If you install the latest version of node and run:

git clone https://github.com/reorproject/reor
npm install
npm run dev

that would be much appreciated :)

i do it,but it tips "Indexing error: TypeError: fetch failed. Please try restarting or open a Github issue.

",and the cmd show

Failed to initialize pipeline TypeError: fetch failed at Object.fetch (node:internal/deps/undici/undici:12293:11) at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at async getModelFile (file:///*****/@[email protected]/node_modules/@xenova/transformers/src/utils/hub.js:471:24) at async getModelJSON (file:///*****/@[email protected]/node_modules/@xenova/transformers/src/utils/hub.js:575:18) at async Promise.all (index 0) at async loadTokenizer (file:///*****/@[email protected]/node_modules/@xenova/transformers/src/tokenizers.js:61:18) at async AutoTokenizer.from_pretrained (file:///*****/@[email protected]/node_modules/@xenova/transformers/src/tokenizers.js:4343:50) at async Promise.all (index 0) at async loadItems (file:///*****/@[email protected]/node_modules/@xenova/transformers/src/pipelines.js:3116:5) at async pipeline (file:///*****/@[email protected]/node_modules/@xenova/transformers/src/pipelines.js:3056:21) { cause: Error: read ECONNRESET at TLSWrap.onStreamRead (node:internal/stream_base_commons:217:20) { errno: -4077, code: 'ECONNRESET', syscall: 'read' } } Error during file indexing: TypeError: fetch failed at Object.fetch (node:internal/deps/undici/undici:12293:11) at process.processTicksAndRejections (node:internal/process/task_queues:95:5)

[gerrampard]

Would you be able to run from source and send me what shows up in logs, just I don't have much to debug with right now. If you install the latest version of node and run:

git clone https://github.com/reorproject/reor
npm install
npm run dev

that would be much appreciated :)

thanks, indexing ok now, but new error: error saving current file! (when i switch from my markdown files on the menu left?? and APP crashed when I submit anything to ASK

[GuoJiaxing1997]

same error for me

[samlhuillier]

I suspect this is related to network settings, proxy/vpn/firewall. Could you guys let me know whether you have any of these and which country you are in etc. I managed to reproduce this by opening a vpn in a different country.

I am working on a patch to make it easier to manually download and add the embedding model if this happens, but would like some more information to solve the root problem.

[samlhuillier]

A new version is out that allows you to download an embedding model from huggingface and attach it yourself to get around this proxy issue with downloading embedding models from huggingface!

And a guide to doing this is available here.

I am also working on a fix to get around the proxy issues as we speak :)

[gerrampard]

A new version is out that allows you to download an embedding model from huggingface and attach it yourself to get around this proxy issue with downloading embedding models from huggingface!

And a guide to doing this is available here.

I am also working on a fix to get around the proxy issues as we speak :)

Initials indexing and model loading seems ok for me, but APP crash every time when I ASK, i've submmited the crash report to Apple, could you please take a look, thanks. (Apple M1 pro mbp, MacOS 14.3.1, Reor 0.1.49)

[samlhuillier]

Thank you @gerrampard for mentioning.

• Which local model, which context length and how much RAM do you have?

Unfortunately, local models are still fairly new so getting the conditions completely right for avoiding crashes is quite hard :(

Also, did attaching a local embedding model and that fixed the fetch error?

[gerrampard]

Thank you @gerrampard for mentioning.

• Which local model, which context length and how much RAM do you have?

Unfortunately, local models are still fairly new so getting the conditions completely right for avoiding crashes is quite hard :(

Also, did attaching a local embedding model and that fixed the fetch error?

It appears that the crash issue was caused by the model I initially selected (chinese-alpaca-2-7b.Q4_K_M.gguf). The issue is now resolved after switching to a new model (openhermes-2.5-mistral-7b.Q6_K.gguf).

Thank you for this amazing tool and your excellent work. I am looking forward to more features in the future!

p.s. The fetch error?No errors for now. fantastic!

[samlhuillier]

Great @gerrampard !

I have now pushed out a fix for the proxy issues which should fix this error message entirely, download the model automatically and get rid of this error: https://github.com/reorproject/reor/releases/tag/v0.1.50 @SYLJX @imb4dm4n @kankanai could you try it and confirm that it works?