Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-21 01:25:28 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #5185] [BUG]: CUDA Tinyscribe transcription fails with TypeError reading 'map' (missing segments) #4955

New Issue
Open
opened 2026-06-05 14:51:06 -04:00 by yindo · 5 comments
yindo commented 2026-06-05 14:51:06 -04:00
Owner
Copy Link
Copy Source

Originally created by @tibacher on GitHub (Mar 10, 2026).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5185

How are you running AnythingLLM?

AnythingLLM desktop app

What happened?

Bug Report: GPU (CUDA) Transcription Fails in Meeting Assistant

Summary

When using AnythingLLM Desktop – Meeting Assistant with the GPU/CUDA transcription engine, all meeting transcriptions fail.
The process starts, runs a few seconds, then aborts.

  • New meetings are disapearing after the failed transcription.
  • Existing meetings remain unchanged (re‑transcription does not update them).
  • Using the CPU engine works 100% reliably.

Steps to Reproduce

  1. Open AnythingLLM Desktop.
  2. Go to Meeting Assistant.
  3. Choose either:
    • Start a new meeting and let it auto‑transcribe, or
    • Open an existing meeting and click Retranscribe.
  4. Ensure that in the engine folder, the CUDA engine is active
    (windows-x64-cuda/tinyscribe.exe).
  5. Start transcription.

Expected Result

The meeting audio is transcribed successfully, and a master-recording.json file is generated for the meeting.

Actual Result

  • Transcription fails within a few seconds.
  • The UI stops the process silently (only “transcription failed”).
  • New meetings are discarded after failure.
  • Existing meetings are not updated.
  • No transcription JSON file is created.

Technical Findings (from logs)

The failure always occurs only when the CUDA engine is used.
The CPU engine works perfectly.

From anonymized logs:

Running Tinyscribe (CUDA) with model: parakeet-tdt-0.6b-v3-onnx
Checking server health... OK
Speaker identification enabled - using /inference

[Tinyscribe] pyannote-rs: Using CUDA execution provider
[Tinyscribe] pyannote-rs: Using CPU execution provider   <-- unexpected fallback

Transcription failed! TypeError: Cannot read properties of undefined (reading 'map')
    at postProcessTranscriptionResponse (.../main/index.js:39:32091)

Error: Master recording transcription file does not exist after conversion!

Interpretation

  • The CUDA engine starts but falls back to CPU internally, which seems unintended. [rsrg-my.sh...epoint.com]
  • Tinyscribe then returns a malformed or incomplete JSON response.
  • The AnythingLLM Electron main process crashes while running:
    response.segments.map(...)
    because segments is undefined.
  • Because of this crash, no transcription JSON is written, and the UI shows a failed transcription.
  • The CPU engine (windows-x64/tinyscribe.exe) always produces a valid response and works without errors.

Impact

  • GPU mode is completely unusable for meeting transcription.
  • Users cannot transcribe or retranscribe any meetings while GPU is active.
  • Workflow interruption: new meetings get deleted after failed transcription.

Suggested Areas to Investigate

  1. Tinyscribe CUDA build

    • Why does CUDA init fail and switch to CPU unexpectedly?
    • Why does the CUDA variant return incomplete JSON (missing segments)?

  2. Electron post‑processor

    • Add schema validation before calling .map(...)
    • Return a recoverable error instead of crashing the transcription job.

Are there known steps to reproduce?

  1. Open AnythingLLM Desktop.
  2. Go to Meeting Assistant.
  3. Choose either:
    • Start a new meeting and let it auto‑transcribe, or
    • Open an existing meeting and click Retranscribe.
  4. Ensure that in the engine folder, the CUDA engine is active
    (windows-x64-cuda/tinyscribe.exe).
  5. Start transcription.
Originally created by @tibacher on GitHub (Mar 10, 2026). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5185 ### How are you running AnythingLLM? AnythingLLM desktop app ### What happened? # **Bug Report: GPU (CUDA) Transcription Fails in Meeting Assistant** ## **Summary** When using **AnythingLLM Desktop – Meeting Assistant** with the **GPU/CUDA transcription engine**, all meeting transcriptions fail. The process starts, runs a few seconds, then aborts. * New meetings are **disapearing** after the failed transcription. * Existing meetings remain **unchanged** (re‑transcription does not update them). * Using the **CPU engine** works **100% reliably**. *** ## **Steps to Reproduce** 1. Open **AnythingLLM Desktop**. 2. Go to **Meeting Assistant**. 3. Choose either: * **Start a new meeting** and let it auto‑transcribe, **or** * Open an existing meeting and click **Retranscribe**. 4. Ensure that in the engine folder, the **CUDA engine** is active (`windows-x64-cuda/tinyscribe.exe`). 5. Start transcription. *** ## **Expected Result** The meeting audio is transcribed successfully, and a `master-recording.json` file is generated for the meeting. *** ## **Actual Result** * Transcription **fails within a few seconds**. * The UI stops the process silently (only “transcription failed”). * New meetings are **discarded** after failure. * Existing meetings are **not updated**. * No transcription JSON file is created. *** ## **Technical Findings (from logs)** The failure always occurs **only when the CUDA engine is used**. The CPU engine works perfectly. From anonymized logs: Running Tinyscribe (CUDA) with model: parakeet-tdt-0.6b-v3-onnx Checking server health... OK Speaker identification enabled - using /inference [Tinyscribe] pyannote-rs: Using CUDA execution provider [Tinyscribe] pyannote-rs: Using CPU execution provider <-- unexpected fallback Transcription failed! TypeError: Cannot read properties of undefined (reading 'map') at postProcessTranscriptionResponse (.../main/index.js:39:32091) Error: Master recording transcription file does not exist after conversion! ### Interpretation * The CUDA engine starts but **falls back to CPU internally**, which seems unintended. [\[rsrg-my.sh...epoint.com\]](https://rsrg-my.sharepoint.com/personal/tim_bacher_rsrg_com/Documents/Microsoft%20Copilot%20Chat-Dateien/meeting-assistant-2026-03-10.log) * Tinyscribe then returns a **malformed or incomplete JSON response**. * The AnythingLLM Electron main process crashes while running: response.segments.map(...) because `segments` is **undefined**. * Because of this crash, **no transcription JSON** is written, and the UI shows a failed transcription. * The CPU engine (`windows-x64/tinyscribe.exe`) always produces a valid response and works without errors. *** ## **Impact** * GPU mode is **completely unusable** for meeting transcription. * Users cannot transcribe or retranscribe any meetings while GPU is active. * Workflow interruption: new meetings get deleted after failed transcription. *** ## **Suggested Areas to Investigate** 1. **Tinyscribe CUDA build** * Why does CUDA init fail and switch to CPU unexpectedly? * Why does the CUDA variant return incomplete JSON (missing `segments`)? 2. **Electron post‑processor** * Add schema validation before calling `.map(...)` * Return a recoverable error instead of crashing the transcription job. ### Are there known steps to reproduce? 1. Open **AnythingLLM Desktop**. 2. Go to **Meeting Assistant**. 3. Choose either: * **Start a new meeting** and let it auto‑transcribe, **or** * Open an existing meeting and click **Retranscribe**. 4. Ensure that in the engine folder, the **CUDA engine** is active (`windows-x64-cuda/tinyscribe.exe`). 5. Start transcription.
yindo added the possible bugneeds info / can't replicateinvestigating labels 2026-06-05 14:51:07 -04:00
yindo commented 2026-06-05 14:51:07 -04:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Mar 10, 2026):

What kind of hardware are you on?

The CUDA engine starts but falls back to CPU internally,

Is intentional, if the GPU cannot be used the CPU process should indeed be the failover.

Can you also post the meeting-assistant logfile in the storage for more descriptive errors?
https://docs.anythingllm.com/installation-desktop/storage#where-is-my-data-located

<!-- gh-comment-id:4031965030 --> @timothycarambat commented on GitHub (Mar 10, 2026): What kind of hardware are you on? > The CUDA engine starts but falls back to CPU internally, Is intentional, if the GPU cannot be used the CPU process should indeed be the failover. Can you also post the meeting-assistant logfile in the storage for more descriptive errors? https://docs.anythingllm.com/installation-desktop/storage#where-is-my-data-located
yindo commented 2026-06-05 14:51:07 -04:00
Author
Owner
Copy Link
Copy Source

@tibacher commented on GitHub (Mar 10, 2026):

I'm running into a reproducible issue where the Meeting Assistant fails to transcribe when using GPU runtime + Speaker Identification. With CPU only set, it works without any errors.

@timothycarambat

Hardware:

  • NVIDIA GPU (CUDA initialized correctly) - NVIDIA RTX PRO 500 Blackwell
  • Windows 11
  • AnythingLLM Desktop latest release - 1.11.1

Below is the full meeting‑assistant log from the session:

{"level":"info","message":"Initialized & Registered","service":"meeting-assistant"}
{"level":"info","message":"FFMPEG bootstrapped successfully!","service":"meeting-assistant"}
{"level":"info","message":"Getting system performance rating...","service":"meeting-assistant"}
{"level":"info","message":"cbe7d9c-8963-4d36-9f40-42fbb8e73b1a - Converting C:\\Users\\XXXX\\Downloads\\Mar 9, 2026 at 12_59 PM Recording.wav to WAV file for transcription...","service":"meeting-assistant"}
{"level":"info","message":"cbe7d9c-8963-4d36-9f40-42fbb8e73b1a - Transcribing master recording file...","service":"meeting-assistant"}
{"level":"info","message":"Looking for master recording file in: C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\meeting-assistant\\cbe7d9c-8963-4d36-9f40-42fbb8e73b1a\\master-recording.wav","service":"meeting-assistant"}
{"level":"info","message":"Retranscribing existing file: C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\meeting-assistant\\cbe7d9c-8963-4d36-9f40-42fbb8e73b1a\\master-recording.wav","service":"meeting-assistant"}
{"level":"info","message":"Expecting Tinyscribe server executable to be at: C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\engines\\meeting-assistant\\windows-x64-cuda\\tinyscribe.exe","service":"meeting-assistant"}
{"level":"info","message":"Running Tinyscribe with model: parakeet-tdt-0.6b-v3-onnx","service":"meeting-assistant"}
{"level":"info","message":"Tinyscribe starting with arguments: --ffmpeg C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\engines\\ffmpeg\\windows-x64 -d C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\models\\meeting-assistant --port 9090","service":"meeting-assistant"}
{"level":"info","message":"Checking server health (1/25)","service":"meeting-assistant"}
{"level":"info","message":"Checking server health (2/25)","service":"meeting-assistant"}
{"level":"info","message":"Server health check passed - service is now online after 2 attempts!","service":"meeting-assistant"}
{"level":"info","message":"Speaker identification enabled - using /inference endpoint","service":"meeting-assistant"}
{"level":"info","message":"[Tinyscribe] stderr [INFO] pyannote-rs: Using CUDA execution provider (NVIDIA GPU)\n","service":"meeting-assistant"}
{"level":"info","message":"[Tinyscribe] stderr [INFO] pyannote-rs: Using CPU execution provider\n","service":"meeting-assistant"}
{"level":"info","message":"master-recording.wav - Transcription failed! TypeError: Cannot read properties of undefined (reading 'map')\n    at e.postProcessTranscriptionResponse (C:\\Users\\XXXX\\AppData\\Local\\Programs\\AnythingLLM\\resources\\app.asar\\dist-electron\\main\\index.js:39:32091)\n    at C:\\Users\\XXXX\\AppData\\Local\\Programs\\AnythingLLM\\resources\\app.asar\\dist-electron\\main\\index.js:39:29768\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)","service":"meeting-assistant"}
{"level":"info","message":"Error transcribing WAV file: null","service":"meeting-assistant"}
{"level":"info","message":"Tinyscribe server process killed as it was started for this re-transcription.","service":"meeting-assistant"}
{"level":"info","message":"Error converting recording to meeting transcript: Error: Master recording transcription file does not exist after conversion!\n    at e.handleConvertRecordingToMeetingTranscript (C:\\Users\\XXXX\\AppData\\Local\\Programs\\AnythingLLM\\resources\\app.asar\\dist-electron\\main\\index.js:39:33539)\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n    at async WebContents.<anonymous> (node:electron/js2c/browser_init:2:83724)","service":"meeting-assistant"}
{"level":"info","message":"Tinyscribe process closed with code: 0","service":"meeting-assistant"}
<!-- gh-comment-id:4033573374 --> @tibacher commented on GitHub (Mar 10, 2026): I'm running into a reproducible issue where the Meeting Assistant fails to transcribe when using **GPU runtime + Speaker Identification**. With CPU only set, it works without any errors. @timothycarambat **Hardware:** * NVIDIA GPU (CUDA initialized correctly) - NVIDIA RTX PRO 500 Blackwell * Windows 11 * AnythingLLM Desktop latest release - 1.11.1 Below is the full meeting‑assistant log from the session: ``` {"level":"info","message":"Initialized & Registered","service":"meeting-assistant"} {"level":"info","message":"FFMPEG bootstrapped successfully!","service":"meeting-assistant"} {"level":"info","message":"Getting system performance rating...","service":"meeting-assistant"} {"level":"info","message":"cbe7d9c-8963-4d36-9f40-42fbb8e73b1a - Converting C:\\Users\\XXXX\\Downloads\\Mar 9, 2026 at 12_59 PM Recording.wav to WAV file for transcription...","service":"meeting-assistant"} {"level":"info","message":"cbe7d9c-8963-4d36-9f40-42fbb8e73b1a - Transcribing master recording file...","service":"meeting-assistant"} {"level":"info","message":"Looking for master recording file in: C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\meeting-assistant\\cbe7d9c-8963-4d36-9f40-42fbb8e73b1a\\master-recording.wav","service":"meeting-assistant"} {"level":"info","message":"Retranscribing existing file: C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\meeting-assistant\\cbe7d9c-8963-4d36-9f40-42fbb8e73b1a\\master-recording.wav","service":"meeting-assistant"} {"level":"info","message":"Expecting Tinyscribe server executable to be at: C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\engines\\meeting-assistant\\windows-x64-cuda\\tinyscribe.exe","service":"meeting-assistant"} {"level":"info","message":"Running Tinyscribe with model: parakeet-tdt-0.6b-v3-onnx","service":"meeting-assistant"} {"level":"info","message":"Tinyscribe starting with arguments: --ffmpeg C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\engines\\ffmpeg\\windows-x64 -d C:\\Users\\XXXX\\AppData\\Roaming\\anythingllm-desktop\\storage\\models\\meeting-assistant --port 9090","service":"meeting-assistant"} {"level":"info","message":"Checking server health (1/25)","service":"meeting-assistant"} {"level":"info","message":"Checking server health (2/25)","service":"meeting-assistant"} {"level":"info","message":"Server health check passed - service is now online after 2 attempts!","service":"meeting-assistant"} {"level":"info","message":"Speaker identification enabled - using /inference endpoint","service":"meeting-assistant"} {"level":"info","message":"[Tinyscribe] stderr [INFO] pyannote-rs: Using CUDA execution provider (NVIDIA GPU)\n","service":"meeting-assistant"} {"level":"info","message":"[Tinyscribe] stderr [INFO] pyannote-rs: Using CPU execution provider\n","service":"meeting-assistant"} {"level":"info","message":"master-recording.wav - Transcription failed! TypeError: Cannot read properties of undefined (reading 'map')\n at e.postProcessTranscriptionResponse (C:\\Users\\XXXX\\AppData\\Local\\Programs\\AnythingLLM\\resources\\app.asar\\dist-electron\\main\\index.js:39:32091)\n at C:\\Users\\XXXX\\AppData\\Local\\Programs\\AnythingLLM\\resources\\app.asar\\dist-electron\\main\\index.js:39:29768\n at process.processTicksAndRejections (node:internal/process/task_queues:95:5)","service":"meeting-assistant"} {"level":"info","message":"Error transcribing WAV file: null","service":"meeting-assistant"} {"level":"info","message":"Tinyscribe server process killed as it was started for this re-transcription.","service":"meeting-assistant"} {"level":"info","message":"Error converting recording to meeting transcript: Error: Master recording transcription file does not exist after conversion!\n at e.handleConvertRecordingToMeetingTranscript (C:\\Users\\XXXX\\AppData\\Local\\Programs\\AnythingLLM\\resources\\app.asar\\dist-electron\\main\\index.js:39:33539)\n at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n at async WebContents.<anonymous> (node:electron/js2c/browser_init:2:83724)","service":"meeting-assistant"} {"level":"info","message":"Tinyscribe process closed with code: 0","service":"meeting-assistant"} ```
yindo commented 2026-06-05 14:51:07 -04:00
Author
Owner
Copy Link
Copy Source

@tibacher commented on GitHub (Mar 13, 2026):

Main issue: The failover to CPU triggers an error, which aborts the transcription process.

When running in CPU-only mode, transcription works without any issues (see logs above for details).

@timothycarambat please reopen the issue.

<!-- gh-comment-id:4053030179 --> @tibacher commented on GitHub (Mar 13, 2026): Main issue: The failover to CPU triggers an error, which aborts the transcription process. When running in CPU-only mode, transcription works without any issues (see logs above for details). @timothycarambat please reopen the issue.
yindo commented 2026-06-05 14:51:08 -04:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Mar 17, 2026):

@tibacher Okay I think I got to the bottom of this. I think the model failed to download all the required onnx files needed which resulted in this failure.

Can you tell me what is in ~/Library/Application Support/anythingllm-desktop/storage/models/meeting-assistant/?

ls -la ~/Library/Application\ Support/anythingllm-desktop/storage/models/meeting-assistant/
total 63520
drwxr-xr-x@ 5 tim  staff       160 Mar 17 11:53 .
drwxr-xr-x@ 7 tim  staff       224 Mar 17 14:07 ..
drwxr-xr-x@ 6 tim  staff       192 Mar 17 11:54 parakeet-tdt-0.6b-v3-onnx
-rw-r--r--@ 1 tim  staff   5983836 Mar 17 11:53 segmentation-3.0.onnx
-rw-r--r--@ 1 tim  staff  26535549 Mar 17 11:53 wespeaker-voxceleb-resnet34-LM.onnx

Then parakeet-tdt-0.6b-v3-onnx should have these:

ls -la ~/Library/Application\ Support/anythingllm-desktop/storage/models/meeting-assistant/parakeet-tdt-0.6b-v3-onnx 
total 4980096
drwxr-xr-x@ 6 tim  staff         192 Mar 17 11:54 .
drwxr-xr-x@ 5 tim  staff         160 Mar 17 11:53 ..
-rw-r--r--@ 1 tim  staff    72520893 Mar 17 11:53 decoder_joint-model.onnx
-rw-r--r--@ 1 tim  staff    41770866 Mar 17 11:53 encoder-model.onnx
-rw-r--r--@ 1 tim  staff  2435420160 Mar 17 11:54 encoder-model.onnx.data
-rw-r--r--@ 1 tim  staff       93939 Mar 17 11:53 vocab.txt

Will do better error reporting for this though and handle that crash case

<!-- gh-comment-id:4078037184 --> @timothycarambat commented on GitHub (Mar 17, 2026): @tibacher Okay I think I got to the bottom of this. I think the model failed to download all the required onnx files needed which resulted in this failure. Can you tell me what is in `~/Library/Application Support/anythingllm-desktop/storage/models/meeting-assistant/`? ``` ls -la ~/Library/Application\ Support/anythingllm-desktop/storage/models/meeting-assistant/ total 63520 drwxr-xr-x@ 5 tim staff 160 Mar 17 11:53 . drwxr-xr-x@ 7 tim staff 224 Mar 17 14:07 .. drwxr-xr-x@ 6 tim staff 192 Mar 17 11:54 parakeet-tdt-0.6b-v3-onnx -rw-r--r--@ 1 tim staff 5983836 Mar 17 11:53 segmentation-3.0.onnx -rw-r--r--@ 1 tim staff 26535549 Mar 17 11:53 wespeaker-voxceleb-resnet34-LM.onnx ``` Then `parakeet-tdt-0.6b-v3-onnx` should have these: ``` ls -la ~/Library/Application\ Support/anythingllm-desktop/storage/models/meeting-assistant/parakeet-tdt-0.6b-v3-onnx total 4980096 drwxr-xr-x@ 6 tim staff 192 Mar 17 11:54 . drwxr-xr-x@ 5 tim staff 160 Mar 17 11:53 .. -rw-r--r--@ 1 tim staff 72520893 Mar 17 11:53 decoder_joint-model.onnx -rw-r--r--@ 1 tim staff 41770866 Mar 17 11:53 encoder-model.onnx -rw-r--r--@ 1 tim staff 2435420160 Mar 17 11:54 encoder-model.onnx.data -rw-r--r--@ 1 tim staff 93939 Mar 17 11:53 vocab.txt ``` Will do better error reporting for this though and handle that crash case
yindo commented 2026-06-05 14:51:08 -04:00
Author
Owner
Copy Link
Copy Source

@tibacher commented on GitHub (Mar 19, 2026):

Here is the Output:

ls C:\Users\XXXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant\

    Directory: C:\Users\XXXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant

Mode                 LastWriteTime         Length Name
----                 -------------         ------ ----
d----          10.03.2026    10:38                .cache
d----          08.03.2026    19:35                parakeet-tdt-0.6b-v3-onnx
-a---          08.03.2026    19:33        5983836 segmentation-3.0.onnx
-a---          08.03.2026    19:33       26535549 wespeaker-voxceleb-resnet34-LM.onnx


> ls C:\Users\XXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant\parakeet-tdt-0.6b-v3-onnx\

    Directory:
C:\Users\XXXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant\parakeet-tdt-0.6b-v3-onnx

Mode                 LastWriteTime         Length Name
----                 -------------         ------ ----
-a---          08.03.2026    19:35       72520893 decoder_joint-model.onnx
-a---          08.03.2026    19:35       41770866 encoder-model.onnx
-a---          08.03.2026    19:39     2435420160 encoder-model.onnx.data
-a---          08.03.2026    19:35          93939 vocab.txt

And for the engine:

> ls C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines\

    Directory: C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines

Mode                 LastWriteTime         Length Name
----                 -------------         ------ ----
d----          08.03.2026    19:33                ffmpeg
d----          10.03.2026    10:47                meeting-assistant
d----          08.03.2026    19:31                ollama


> ls C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines\meeting-assistant\

    Directory: C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines\meeting-assistant

Mode                 LastWriteTime         Length Name
----                 -------------         ------ ----
d----          08.03.2026    19:33                windows-x64
d----          10.03.2026    10:47                windows-x64-cuda
-a---          08.03.2026    19:33              5 .meeting-assistant-bootstrap-complete
<!-- gh-comment-id:4088275140 --> @tibacher commented on GitHub (Mar 19, 2026): Here is the Output: ```pwsh ls C:\Users\XXXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant\ Directory: C:\Users\XXXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant Mode LastWriteTime Length Name ---- ------------- ------ ---- d---- 10.03.2026 10:38 .cache d---- 08.03.2026 19:35 parakeet-tdt-0.6b-v3-onnx -a--- 08.03.2026 19:33 5983836 segmentation-3.0.onnx -a--- 08.03.2026 19:33 26535549 wespeaker-voxceleb-resnet34-LM.onnx > ls C:\Users\XXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant\parakeet-tdt-0.6b-v3-onnx\ Directory: C:\Users\XXXX\AppData\Roaming\anythingllm-desktop\storage\models\meeting-assistant\parakeet-tdt-0.6b-v3-onnx Mode LastWriteTime Length Name ---- ------------- ------ ---- -a--- 08.03.2026 19:35 72520893 decoder_joint-model.onnx -a--- 08.03.2026 19:35 41770866 encoder-model.onnx -a--- 08.03.2026 19:39 2435420160 encoder-model.onnx.data -a--- 08.03.2026 19:35 93939 vocab.txt ```` And for the engine: ```pwsh > ls C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines\ Directory: C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines Mode LastWriteTime Length Name ---- ------------- ------ ---- d---- 08.03.2026 19:33 ffmpeg d---- 10.03.2026 10:47 meeting-assistant d---- 08.03.2026 19:31 ollama > ls C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines\meeting-assistant\ Directory: C:\Users\xxxx\AppData\Roaming\anythingllm-desktop\storage\engines\meeting-assistant Mode LastWriteTime Length Name ---- ------------- ------ ---- d---- 08.03.2026 19:33 windows-x64 d---- 10.03.2026 10:47 windows-x64-cuda -a--- 08.03.2026 19:33 5 .meeting-assistant-bootstrap-complete ```
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 investigating needs info / can't replicate 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#4955
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?