Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-19 14:13:35 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #3161] [BUG]: Context window size for TogetherAI Deepseek-chat #2029

New Issue
Closed
opened 2026-02-22 18:27:49 -05:00 by yindo · 1 comment
yindo commented 2026-02-22 18:27:49 -05:00
Owner
Copy Link
Copy Source

Originally created by @Taluen79 on GitHub (Feb 9, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3161

How are you running AnythingLLM?

AnythingLLM desktop app

What happened?

You guys must have butchered the context behind the scene somehow. Because I was talking to DeepSeek V3 via TogetherAI, and about 15 short messages into the conversation, it started to reply inconsistently, as if it does not have any idea what I was talking about.

Are there known steps to reproduce?

The chat was very simple. I gave it some rule to pronounce different words and then when I provide one word, it would give me the pronunciation. After just a 4-5 words, I asked which words we have gone thru so far, and it can only remember 1 of the words. Even when I tried to hint that perhaps we have talked more than 1 word, it still insisted that we have only gone thru 1 word. And this word was not even the most recent one, but something earlier.

I did the same thing in Open WebUI, using the same model and provider and it worked flawlessly for many words. And it can even give me a complete list of all the words that we have discussed.

Originally created by @Taluen79 on GitHub (Feb 9, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3161 ### How are you running AnythingLLM? AnythingLLM desktop app ### What happened? You guys must have butchered the context behind the scene somehow. Because I was talking to DeepSeek V3 via TogetherAI, and about 15 short messages into the conversation, it started to reply inconsistently, as if it does not have any idea what I was talking about. ### Are there known steps to reproduce? The chat was very simple. I gave it some rule to pronounce different words and then when I provide one word, it would give me the pronunciation. After just a 4-5 words, I asked which words we have gone thru so far, and it can only remember 1 of the words. Even when I tried to hint that perhaps we have talked more than 1 word, it still insisted that we have only gone thru 1 word. And this word was not even the most recent one, but something earlier. I did the same thing in Open WebUI, using the same model and provider and it worked flawlessly for many words. And it can even give me a complete list of all the words that we have discussed.
yindo added the possible bug label 2026-02-22 18:27:49 -05:00
yindo closed this issue 2026-02-22 18:27:50 -05:00
yindo commented 2026-02-22 18:27:50 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Feb 10, 2025):

The issue here is almost certainly the context length / maxLength.

If you open the desktop app storage folder and then go to models > togetherai you should see two files models.json and .cached_at. Can you paste or attach models.json.

There should be an entry in there for Deepseek-Chat or whatever the model key is on TogetherAI and it should have a property of maxLength that should be the max of 128,000 like the normal model has.

@timothycarambat commented on GitHub (Feb 10, 2025): The issue here is almost certainly the context length / `maxLength`. If you open the [desktop app storage folder](https://docs.anythingllm.com/installation-desktop/storage) and then go to `models > togetherai` you should see two files `models.json` and `.cached_at`. Can you paste or attach `models.json`. There should be an entry in there for `Deepseek-Chat` or whatever the model key is on TogetherAI and it should have a property of `maxLength` that should be the max of `128,000` like the normal model has.
yindo changed title from [BUG]: Context window size for TogetherAI Deepseek-chat to [GH-ISSUE #3161] [BUG]: Context window size for TogetherAI Deepseek-chat 2026-06-05 14:44:07 -04:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label possible bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2029
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?