Louis
2e9b7fdad2
chore: add import name for legacy models
2024-11-07 13:34:09 +07:00
Louis
40019892b8
chore: correct name of bin subfolders to move dll properly
2024-11-07 13:01:34 +07:00
Louis
ced44973b8
chore: queue server start and model load
2024-11-07 12:06:46 +07:00
Louis
e41bcffcef
fix: export PATH env to engine destination folder to have additional dlls scoped
2024-11-07 10:10:05 +07:00
Louis
264720c71a
chore: support customized OpenAI model.json
2024-11-06 16:46:27 +07:00
Louis
56e35df84d
chore: clean dangling process on exit and relaunch
2024-11-06 13:34:11 +07:00
Louis
46d5faf59f
chore: new cortex-cpp binary - model import option and model size
2024-11-04 20:36:04 +07:00
Louis
a986c6de2d
chore: decide model name on pull and import
2024-11-04 15:37:20 +07:00
Louis
5ddbf5fb34
fix: unlink the entire model folder on delete
2024-11-04 15:37:20 +07:00
Louis
1ab02b706f
fix: model import symlink
2024-11-04 15:37:19 +07:00
Louis
61f72e6775
chore: bump cortex-cpp v1.0.2-rc1
2024-11-04 15:37:19 +07:00
Louis
e5f5d887e3
fix: persists model.json on download (legacy models)
2024-11-04 15:37:19 +07:00
Louis
a466bbca38
chore: update legacy tensorrt-llm download and run
2024-11-04 15:37:19 +07:00
Louis
2c11caf87e
chore: shared cuda dependencies
2024-11-04 15:37:18 +07:00
Louis
3643c8866e
fix: correct model settings on startup and strip down irrelevant model parameters
2024-11-04 15:37:18 +07:00
Louis
8f778ee90f
feat: app supports cortex.cpp model downloader and legacy downloader - maintain legacy JSON models
2024-11-04 15:37:18 +07:00
Louis
5f075c8554
fix: prebundle cudart and cublas
2024-11-04 15:37:18 +07:00
Louis
dc87f37a9b
fix: package cortex.cpp engines and cuda on windows
2024-11-04 15:37:17 +07:00
Louis
a0e2f16a3b
chore: binary naming convention - following llama.cpp release
2024-11-04 15:37:17 +07:00
Louis
03333cc4c2
fix: onboarding should cover cortex models - debounce reduce model reload - rename cortex binary name
2024-11-04 15:37:17 +07:00
Louis
40957f7686
fix: model reload state - reduce model unload events emit
2024-11-04 15:37:15 +07:00
Louis
523c745150
chore: try catch legacy assistant creation
2024-11-04 15:37:15 +07:00
Louis
716fd96d56
test: add tests for migration strategy
2024-11-04 15:37:15 +07:00
Louis
5edf121d96
test: add tests to legacy model-json utilities
2024-11-04 15:37:15 +07:00
Louis
895c3d4246
fix: tests - useModels with remote models filter
2024-11-04 15:37:15 +07:00
Louis
ba59425e6a
fix: tests
2024-11-04 15:37:14 +07:00
Louis
03e15fb70f
feat: sync model hub and download progress from cortex.cpp
2024-11-04 15:37:14 +07:00
Louis
f44f291bd8
chore: download progress finished should reload model list
2024-11-04 15:37:13 +07:00
Louis
4080dc4b65
feat: model and cortex extensions update
2024-11-04 15:37:12 +07:00
Faisal Amir
b37d4a5c7e
fix: types issue (#internalTypeOnlyBrand) in the @types/node package ( #3921 )
2024-10-31 21:17:25 +07:00
Faisal Amir
267f3ab051
fix: deprecated gpt with vision ( #3912 )
...
* fix: deprecated gpt 4 with vision
* chore: update package version inference openai extension
2024-10-30 17:42:17 +07:00
Louis
53098699ef
Merge pull request #3857 from Haleshot/haleshot/martian-api-hyperlink-fix
...
Update broken/outdated hyperlink
2024-10-23 10:08:28 +07:00
Faisal Amir
b14f54e866
fix: inconsistent state of downloading multimodal ( #3862 )
2024-10-22 15:44:13 +07:00
Srihari Thyagarajan
4c562c3e12
Update broken/outdated hyperlink
2024-10-21 23:53:02 +05:30
Louis
4983247918
fix: correct eos token of llava models
2024-10-21 12:58:18 +07:00
Louis
024992264f
fix: error handling for model imports should be handled gracefully
2024-10-03 19:44:52 +07:00
Louis
6082959d17
chore: update models stop token
2024-10-02 11:46:42 +07:00
Louis
358d8f83f9
chore: add llama3.2 and qwen models
2024-10-02 11:36:55 +07:00
Louis
8334076047
fix : #3491 - Unable to use tensorrt-llm ( #3741 )
...
* fix : #3491 - Unable to use tensorrt-llm
* fix: abortModelDownload input type
2024-09-30 11:58:46 +07:00
Louis
143f2f5c58
fix: wrong model download location when there is a mismatch model_id ( #3733 )
2024-09-26 12:43:34 +07:00
Louis
7f08f0fa79
fix : #3703 - Deepseek-Coder-33B-Instruct is incompatible ( #3732 )
2024-09-26 12:43:23 +07:00
Louis
acd3be3a2a
fix : #3698 - o1 preview models do not work with max_tokens ( #3728 )
2024-09-24 16:35:08 +07:00
Louis
6af17c6455
fix : #3513 - anthropic extension does not forward the system prompt ( #3724 )
2024-09-24 10:40:32 +07:00
Louis
c0b59ece4d
fix : #3558 wrong model metadata import or download from HuggingFace ( #3725 )
...
* fix : #3558 wrong model metadata import
* chore: remove redundant metadata retrieval
2024-09-24 10:07:53 +07:00
Louis
aee8624338
fix : #3693 broken thread.json should not break the entire threads ( #3709 )
...
* fix : #3693 broken thread.json should not break the entire threads
* test: add tests
2024-09-23 14:20:01 +07:00
Louis
8e603bd5db
fix : #3476 - Mismatch id between model json and path ( #3645 )
...
* fix: mismatch between model json and path
* chore: revert preserve model settings
* test: add tests
2024-09-17 16:43:47 +07:00
Louis
c8a08f1115
fix: correct prompt template for Phi3 Medium model ( #3670 )
2024-09-17 09:25:55 +07:00
Ronnie Ghose
670013baa0
Add support for 'o1-preview' and 'o1-mini' models ( #3659 )
...
Add support for 'o1-preview' and 'o1-mini' model names in the OpenAI API.
* **Update `models.json`**:
- Add 'o1-preview' model details with appropriate parameters and metadata.
- Add 'o1-mini' model details with appropriate parameters and metadata.
---
For more details, open the [Copilot Workspace session](https://copilot-workspace.githubnext.com/janhq/jan?shareId=XXXX-XXXX-XXXX-XXXX ).
2024-09-17 09:25:08 +07:00
Louis
98bef7b7cf
test: add model parameter validation rules and persistence tests ( #3618 )
...
* test: add model parameter validation rules and persistence tests
* chore: fix CI cov step
* fix: invalid model settings should fallback to origin value
* test: support fallback integer settings
2024-09-17 08:34:58 +07:00
Louis
ba1ba89fa3
fix: change Phi-3 Mini Instruct 4K to cortexso repository ( #3655 )
...
* fix: change Phi-3 Mini Instruct 4K to cortexso repository
* chore: correct model names
* fix: Phi3-Medium prompt correction
* chore: bump versions
* chore: take out some feature tags
2024-09-13 17:19:00 +07:00
