lilacai / lilac

Curate better data for LLMs

http://lilacml.com

lilacai/lilac Issues

Add `Dataset.load_embedding`
Updated 5 months ago
USE_TABLE_INDEX=True when swapping a space during a reboot causes a lock issue.
Updated 5 months ago
Reduce polling frequency of tasks when no task active
Updated 5 months ago
OpenAI Authentication error when computing embeddings via python API
Closed 5 months ago3
[UI] When searching a ShareGPT dataset, we should allow passing a filter of conversations.*.from= to the search.
Updated 5 months ago
dataset.export() should use combine_cols=True by default.
Updated 5 months ago
UI progress bars lost count, ex/sec, description information.
Closed 5 months ago
Parallel Datasets
Updated 5 months ago7
API Error When Interacting with Dataset Endpoints
Updated 5 months ago1
Error while using OpenAI embeddings
Closed 5 months ago2
Add keyboard shortcuts for fast labeling
Closed 5 months ago1
Add model selection or ability to overwrite GPT model for concept examples generation
Updated 5 months ago1
Environment flags that are boolean still are truthy when you set the string to "False"
Updated 5 months ago
Cannot create concepts in huggingface space
Closed 5 months ago2
Signals and embeddings are much slower
Updated 5 months ago
Rendering bug when we have `a[].b[]`
Closed 5 months ago1
Loading a dataset should not block the REST endpoint.
Closed 5 months ago
Allow datasets with spaces in the name.
Updated 6 months ago
Markdown code block aware chunking.
Updated 6 months ago
ShareGPT format: compare to menu should show the title instead of `conversation.0.value`
Closed 6 months ago3
Add ability to jump to a random integer offset in single-item view.
Updated 6 months ago
Allow user-provided embeddings when ingesting from a source.
Updated 6 months ago
Default to combine_cols=True
Updated 6 months ago
labels stay in ui after deletion.
Updated 6 months ago
Error Computing Embedding with Parquet File in Lilac v0.1.24 on Windows 11
Updated 6 months ago
Add `dataset.map(embeddings=True)`
Updated 6 months ago
Datasets w/ map bug: DataType has no attribute 'key_type'
Updated 6 months ago
Confused by HDBscan in UI
Updated 6 months ago2
Dockerfile to build application front and back from source
Closed 6 months ago4
Remove num_rows from manifest and add a count() endpoint
Closed 6 months ago1
Add `dataset.count(filters=...)` public API
Updated 6 months ago
duckdb syntax error
Updated 6 months ago
Add `input_cols` to `Dataset.map` to allow an input column to be processed.
Closed 6 months ago2
Improve the CSS for markdown rendered content.
Closed 6 months ago1
Allow subsampling datasets
Closed 6 months ago1
Add documentation in the guide for limit & filter.
Closed 6 months ago
Ergonomics and small bugs with `dataset.map`
Closed 6 months ago2
Multiprocess `map` got ~40% slower
Closed 6 months ago1
website: cross-wired footer follow links
Closed 6 months ago2
I want to modify the Datasets directly
Updated 6 months ago1
Allow deletion of columns created from map from the UI.
Updated 6 months ago1
Multithreaded map only shows progress from shard 0
Closed 6 months ago1
Use a single tqdm for task, instead of showing the sharded tqdms.
Closed 6 months ago
`Map` spans the ipython output cell and freezes the notebook while running
Closed 6 months ago1
Break up the TaskManager.
Updated 6 months ago
Write a getting started guide.
Updated 6 months ago
Build versioning of source & map data.
Updated 6 months ago
Add a simple signal python editor from the UI.
Updated 6 months ago
Compute signal page is broken -- no fields available.
Closed 7 months ago
Add defragmentation of signal output
Updated 7 months ago