Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / TransformerLensOrg/TransformerLens issues and pull requests
#566 - moved report to static section
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#565 - Revert "moved coverage report download (#564)"
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#564 - moved coverage report download
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#563 - Refactor components
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#562 - Ci full coverage
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#561 - Ci coverage location
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#560 - Resolve SAE CI Test failures
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#559 - reworked CI to publish code coverage report
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#558 - [Bug Report] Gemma-2b model fails when splitting QKV input
Issue -
State: closed - Opened by Aaquib111 10 months ago
- 2 comments
#557 - [Question] Difference in gpt-neo-125m weights loading with huggingface from_pretrained and HookTransformers.from_pretrained
Issue -
State: closed - Opened by petezone 10 months ago
- 1 comment
#556 - Bert demo ci
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#555 - removed deuplicate rearrange block
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#554 - [Proposal] Save and Load subgraph as dict
Issue -
State: closed - Opened by chris-aeviator 10 months ago
- 7 comments
#553 - [Bug Report] Cannot load llama-3 8B with hf_model arg
Issue -
State: closed - Opened by Butanium 10 months ago
- 6 comments
#552 - Hf secret
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#551 - Fixed device being set to cpu:0 instead of cpu
Pull Request -
State: closed - Opened by Butanium 10 months ago
- 2 comments
#550 - Hf token auth
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#549 - Add support for Llama 3 (and Llama-2-70b-hf)
Pull Request -
State: closed - Opened by joelburget 10 months ago
- 5 comments
#548 - [Bug Report] Deployments no longer possible without huggingface access token
Issue -
State: closed - Opened by bryce13950 10 months ago
Labels: bug, blocking
#547 - Llama 4bit v2 typing
Pull Request -
State: closed - Opened by hannamw 10 months ago
- 1 comment
#546 - Fixed Santa Coder demo
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#545 - Othello colab fix
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#544 - Demo no position fix
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#543 - [Bug Report] Grokking demo currently broken in Colab
Issue -
State: open - Opened by bryce13950 10 months ago
- 1 comment
Labels: bug, demo
#542 - revised demo testing to check all demos
Pull Request -
State: open - Opened by bryce13950 10 months ago
#541 - locked attribution patching to 1.1.1
Pull Request -
State: closed - Opened by bryce13950 10 months ago
#540 - [Bug Report] Test coverage missing on add_hook in hook_points
Issue -
State: open - Opened by bryce13950 10 months ago
- 2 comments
Labels: bug, testing
#539 - [Bug Report] Transformer_lens should be pinned to 1.1.1 for Attribution_Patching demo
Issue -
State: closed - Opened by adamkarvonen 10 months ago
- 1 comment
#538 - Standardize black line length to 100, in line with other project settings
Pull Request -
State: closed - Opened by Chanlaw 10 months ago
- 4 comments
#537 - Add Xavier and Kaiming Initializations
Pull Request -
State: closed - Opened by Chanlaw 10 months ago
- 3 comments
#536 - HookedSAETransformer
Pull Request -
State: closed - Opened by ckkissane 10 months ago
- 5 comments
#535 - [Proposal] Please ddd support for fine-tuned models
Issue -
State: closed - Opened by jasonlim131 10 months ago
- 3 comments
#534 - Bugfix: remove redundant assert checks
Pull Request -
State: closed - Opened by tkukurin 11 months ago
- 1 comment
#533 - updated docs to account for additional test suites
Pull Request -
State: closed - Opened by bryce13950 11 months ago
#530 - [Bug Report] Demos utilizing PySvelte are not able to install
Issue -
State: closed - Opened by vashchuko 11 months ago
- 3 comments
Labels: demo
#529 - Update loading_from_pretrained.py
Pull Request -
State: closed - Opened by jbloomAus 11 months ago
- 2 comments
#528 - make tests pass mps
Pull Request -
State: closed - Opened by jbloomAus 11 months ago
#524 - Remove FactoredMatrix.py<->utils.py circular dependency
Pull Request -
State: open - Opened by ArthurConmy 11 months ago
#523 - [Bug Report] Residual Stack Not Adding Up
Issue -
State: open - Opened by EitanGronich 11 months ago
- 1 comment
Labels: documentation, demo
#522 - [Proposal] Include num_kv_heads in the Model Properties Table in the docs
Issue -
State: closed - Opened by neelnanda-io 11 months ago
- 2 comments
#521 - Add Mixtral
Pull Request -
State: closed - Opened by collingray 11 months ago
- 12 comments
#520 - Fix split_qkv_input for grouped query attention
Pull Request -
State: closed - Opened by wesg52 11 months ago
- 1 comment
#519 - [Bug Report] split_qkv_input issues with consistency and GQA
Issue -
State: closed - Opened by wesg52 11 months ago
- 1 comment
#518 - [Proposal] Update Attention parameter initialization to use Kaiming
Issue -
State: open - Opened by justjhong 11 months ago
Labels: complexity-moderate
#517 - [Bug Report] Issues loading Llama-2
Issue -
State: closed - Opened by davidquarel 12 months ago
- 2 comments
#516 - chore: fixing type errors and enabling mypy
Pull Request -
State: closed - Opened by chanind 12 months ago
- 4 comments
#515 - In evals.IOIDataset, all entries in the dataset are the same.
Issue -
State: open - Opened by tkwa 12 months ago
Labels: low-priority, complexity-simple
#510 - Speed up !pip install transformer-lens in colab
Pull Request -
State: closed - Opened by pavanyellow 12 months ago
- 3 comments
#509 - [Bug Report] Layer norm folding not properly implemented for BertBlock
Issue -
State: open - Opened by soniajoseph about 1 year ago
- 2 comments
Labels: bug, complexity-moderate
#506 - [Proposal] Add support for Qwen1.5 models
Issue -
State: closed - Opened by andyrdt about 1 year ago
- 1 comment
#505 - Refactor hook_points
Pull Request -
State: closed - Opened by VasilGeorgiev39 about 1 year ago
- 4 comments
#502 - [Question] How to migrate and use it on huggingface’s visual language model?
Issue -
State: open - Opened by jiabao-wang about 1 year ago
- 1 comment
Labels: complexity-high
#501 - [Draft] Support Flash Attention
Pull Request -
State: open - Opened by cmathw about 1 year ago
- 4 comments
#500 - [Question] Generation not possible with hooks?
Issue -
State: closed - Opened by FergusFettes about 1 year ago
- 1 comment
#497 - [Proposal] Speed up `!pip install transformer-lens` in colab
Issue -
State: closed - Opened by ArthurConmy about 1 year ago
- 4 comments
#496 - [Proposal] Add benchmarks for error introduced by HookedTransformer
Issue -
State: closed - Opened by collingray about 1 year ago
#494 - Add Support for Yi-6B and Yi-34B
Pull Request -
State: closed - Opened by collingray about 1 year ago
#493 - Construct causal mask on-the-fly
Pull Request -
State: closed - Opened by andyrdt about 1 year ago
- 8 comments
#492 - [Bug Report] Tiny stories models have longer n_ctx than they were trained with
Issue -
State: open - Opened by nix-apollo about 1 year ago
Labels: good first issue, complexity-simple, implementation-inaccuracy
#488 - add optional arguments to make sure generate() works without tokenizer being defined
Pull Request -
State: closed - Opened by JackCai1206 about 1 year ago
- 1 comment
#486 - Loading of huggingface 4-bit quantized Llama
Pull Request -
State: closed - Opened by coolvision about 1 year ago
- 9 comments
#483 - [Bug Report] HookedTranformer.generate() with model.tokenizer unset gives pad_token_id error
Issue -
State: open - Opened by JackCai1206 about 1 year ago
- 2 comments
Labels: bug, low-priority, complexity-moderate
#482 - [Question] demo of 4bit quantized Llama -- what's next?
Issue -
State: closed - Opened by coolvision about 1 year ago
- 2 comments
#481 - [Bug Report] Demo for Tracr to TransformerLens is kind of broken
Issue -
State: open - Opened by FlyingPumba about 1 year ago
Labels: bug, demo
#480 - [Question] load_state_dict: copy vs assign
Issue -
State: closed - Opened by coolvision about 1 year ago
- 3 comments
#479 - [Proposal] Memory efficient causal mask implementation
Issue -
State: open - Opened by andyrdt about 1 year ago
- 1 comment
Labels: enhancement, complexity-moderate
#474 - [Bug Report] Slack link is expired
Issue -
State: closed - Opened by Aryan-Deshpande about 1 year ago
- 2 comments
#473 - Make `tokenize_and_concatenate` work with more datasets
Pull Request -
State: open - Opened by ArthurConmy about 1 year ago
Labels: enhancement
#472 - [Bug Report] `ActivationCache.apply_ln_to_stack` silently returns the resid stack for unsupported normalization types
Issue -
State: closed - Opened by jmsdao about 1 year ago
- 1 comment
#471 - [Proposal] Support Mixtral
Issue -
State: closed - Opened by neelnanda-io about 1 year ago
- 9 comments
Labels: enhancement, help wanted
#470 - [Bug Report] convert_llama_weights fails if I already quantized the weights to 4 bits
Issue -
State: closed - Opened by abdurraheemali about 1 year ago
- 2 comments
#468 - [Proposal] Add base parameter of rotary embedding in model configuration
Issue -
State: closed - Opened by YuhengHuang42 about 1 year ago
#467 - [Bug Report] Attention masking is not used by model forward methods
Issue -
State: closed - Opened by jmsdao about 1 year ago
- 1 comment
#466 - (Draft) Add DLA function to utils
Pull Request -
State: open - Opened by VasilGeorgiev39 about 1 year ago
- 5 comments
#465 - Add pos_slice to run_with_cache
Pull Request -
State: closed - Opened by VasilGeorgiev39 about 1 year ago
- 3 comments
#464 - Draft PR: add Pyright static typing to hook_points.py
Pull Request -
State: closed - Opened by starship006 about 1 year ago
- 6 comments
#463 - Clean up project config
Pull Request -
State: closed - Opened by alan-cooney about 1 year ago
#462 - [Proposal] Add support for Mamba
Issue -
State: open - Opened by joker3212 about 1 year ago
- 5 comments
Labels: complexity-high, model-request
#460 - [Question] weight decay for layernorms and biases?
Issue -
State: closed - Opened by matthiasdellago about 1 year ago
- 1 comment
#459 - [Question] I tried to load the bert model but got an error
Issue -
State: closed - Opened by clclclaiggg about 1 year ago
- 2 comments
#452 - Refactor components
Pull Request -
State: closed - Opened by bryce13950 about 1 year ago
- 3 comments
#446 - [Bug fix] GatedMLP not in docs. issue #264 [bug report]
Pull Request -
State: closed - Opened by danlaudk about 1 year ago
- 1 comment
#444 - Fix contributing docs typo
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#442 - Relax CUDA requirements
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#440 - add survey link
Pull Request -
State: closed - Opened by jbloomAus over 1 year ago
#439 - [Bug Report] Load model to mutilple devices
Issue -
State: open - Opened by liuxin99 over 1 year ago
- 7 comments
Labels: bug, multi-gpu
#437 - Update GitHub CD Actions
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#436 - Add docs hot reloading instructions for contributors
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#435 - Make unit & acceptance tests run in parallel
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#432 - Improve ActivationCache docs
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#431 - Move cspell conf to its own file
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#430 - Organise & fix README
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
- 2 comments
#429 - Fix Exploratory Analysis Demo
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#428 - Sync readme with docs
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
- 3 comments
#422 - Fix all Sphinx warnings
Pull Request -
State: closed - Opened by alan-cooney over 1 year ago
#385 - [Bug Report] Pythia models / Rotary Embeddings don't match Huggingface.
Issue -
State: open - Opened by UFO-101 over 1 year ago
- 15 comments
Labels: complexity-high, implementation-inaccuracy
#375 - Fix to include ln_final.w in RMSNorm hook
Pull Request -
State: closed - Opened by clarenceluo78 over 1 year ago
- 5 comments
Labels: no-rebase
#327 - Reduce memory use when loading model
Pull Request -
State: closed - Opened by slavachalnev over 1 year ago
- 7 comments
#258 - [Proposal] Support BERT Model Series (and/or T5).
Issue -
State: closed - Opened by jbloomAus almost 2 years ago
- 16 comments
#207 - [Bug Report] Can't add hook to pretrained model: AssertionError: Cannot add hook blocks.0.hook_q_input if use_split_qkv_input is False
Issue -
State: open - Opened by jbloomAus almost 2 years ago
- 4 comments
Labels: bug, help wanted