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
#855 - moved setup python
Pull Request -
State: closed - Opened by bryce13950 4 days ago
#854 - Ci hf secret
Pull Request -
State: closed - Opened by bryce13950 4 days ago
#853 - Ci hf token empty
Pull Request -
State: closed - Opened by bryce13950 4 days ago
#852 - Visualize weight conversions
Pull Request -
State: open - Opened by degenfabian 4 days ago
#851 - Visualize weight conversions
Pull Request -
State: closed - Opened by degenfabian 4 days ago
#850 - updated artifacts version
Pull Request -
State: closed - Opened by bryce13950 5 days ago
#849 - Upgrade transformers
Pull Request -
State: closed - Opened by bryce13950 5 days ago
#848 - Release 2.13.0
Pull Request -
State: closed - Opened by bryce13950 5 days ago
#847 - feat: streaming response for HookedTransformer.generate
Pull Request -
State: open - Opened by hijohnnylin 9 days ago
- 3 comments
#846 - [Bug Report] Prioritize Local hf_model.config for Qwen Models to Avoid Unnecessary Hugging Face API Calls
Issue -
State: open - Opened by yhr-code 9 days ago
Labels: high-priority, complexity-moderate
#845 - Manually create repr for partial hooks
Pull Request -
State: closed - Opened by danbraunai 11 days ago
- 3 comments
#844 - [Compatibility Report] meta-llama/Llama-3.3-70B-Instruct does not work
Issue -
State: open - Opened by thisnick 16 days ago
- 2 comments
Labels: complexity-moderate, implementation-inaccuracy
#843 - added python 3.13 to CI
Pull Request -
State: open - Opened by bryce13950 18 days ago
#842 - [Question] code error when context length increases
Issue -
State: closed - Opened by gom168 18 days ago
#841 - Recent releases
Pull Request -
State: closed - Opened by bryce13950 19 days ago
#840 - Upstream update
Pull Request -
State: closed - Opened by bryce13950 19 days ago
#839 - Phi 4 docs fix
Pull Request -
State: closed - Opened by bryce13950 19 days ago
#838 - Enables return of activation cache variables during generation
Pull Request -
State: open - Opened by japhba 20 days ago
#837 - [Bug Report] Multi-GPU device ordinal issue
Issue -
State: open - Opened by safiness 21 days ago
Labels: bug, complexity-high, multi-gpu
#836 - aded changes needed for phi-4 docs
Pull Request -
State: closed - Opened by bryce13950 21 days ago
#835 - Version 2.12
Pull Request -
State: closed - Opened by bryce13950 21 days ago
#834 - Rotary factory
Pull Request -
State: open - Opened by jonasrohw 24 days ago
#833 - Added model Phi 4
Pull Request -
State: closed - Opened by jonasrohw 24 days ago
#832 - Extend support for T5 models
Pull Request -
State: closed - Opened by degenfabian 24 days ago
#831 - updated lock command
Pull Request -
State: closed - Opened by bryce13950 25 days ago
#830 - [Proposal] Type hint support for `self.model` in `ActivationCache`
Issue -
State: open - Opened by Ja1Zhou 25 days ago
- 1 comment
Labels: complexity-moderate
#829 - Extend Bert support
Pull Request -
State: open - Opened by degenfabian about 1 month ago
- 1 comment
#828 - [Bug Report] Logit attribution produces different values for Gemma2
Issue -
State: closed - Opened by wj210 about 1 month ago
- 1 comment
#827 - Release 2.11
Pull Request -
State: closed - Opened by bryce13950 about 1 month ago
#826 - Feature llama 33
Pull Request -
State: closed - Opened by bryce13950 about 1 month ago
#825 - Depen transformers
Pull Request -
State: closed - Opened by bryce13950 about 1 month ago
- 1 comment
#824 - Jaxtyiping Dependency Update
Pull Request -
State: open - Opened by bryce13950 about 1 month ago
#823 - Test match hugging face
Pull Request -
State: closed - Opened by bryce13950 about 1 month ago
#822 - Updates torch to use the most recent version
Pull Request -
State: closed - Opened by bryce13950 about 1 month ago
#821 - updated python requirements
Pull Request -
State: closed - Opened by bryce13950 about 1 month ago
#820 - Add LLaVA support, modify generate function
Pull Request -
State: closed - Opened by zazamrykh about 2 months ago
- 12 comments
#819 - [Bug Report] Memory Leakage in GEMMA2 -2b
Issue -
State: closed - Opened by stupid-learner about 2 months ago
- 1 comment
#818 - Add LLaVA support and make HookedTransformer.generate() able to get embedding as inputs
Pull Request -
State: closed - Opened by zazamrykh about 2 months ago
- 4 comments
#817 - fixed corner param
Pull Request -
State: closed - Opened by bryce13950 about 2 months ago
#816 - Added OLMo(E) v1
Pull Request -
State: open - Opened by jonasrohw about 2 months ago
#815 - Set prepend_bos to false by default for Qwen models
Pull Request -
State: closed - Opened by degenfabian about 2 months ago
#814 - Fix that padding_side always defaults to "right" when no value is explicitly passed
Pull Request -
State: open - Opened by degenfabian about 2 months ago
- 1 comment
#813 - Release 2.10
Pull Request -
State: closed - Opened by bryce13950 about 2 months ago
#812 - Updated devcontainers to use python3.11
Pull Request -
State: closed - Opened by jonasrohw about 2 months ago
#811 - Add support for Qwen_with_Questions
Pull Request -
State: closed - Opened by degenfabian about 2 months ago
#810 - Throw error when using attn_in with grouped query attention
Pull Request -
State: closed - Opened by degenfabian about 2 months ago
#809 - Added support for Qwen2.5
Pull Request -
State: closed - Opened by israel-adewuyi 2 months ago
#808 - [Proposal] Add support for QwQ models
Issue -
State: closed - Opened by bryce13950 2 months ago
- 1 comment
Labels: complexity-moderate, model-request
#807 - Add a demo of collecting activations from a single location in the model.
Pull Request -
State: closed - Opened by adamkarvonen 2 months ago
- 1 comment
#806 - Set default_prepend_bos to False in Bloom model configuration
Pull Request -
State: closed - Opened by degenfabian 2 months ago
- 2 comments
#805 - [Question] What is stanford-gpt2-small-a?
Issue -
State: closed - Opened by leo1oel 2 months ago
- 1 comment
#804 - [Proposal] Compatibility for OLMo and OLMo2?
Issue -
State: open - Opened by spaidataiga 2 months ago
- 4 comments
Labels: complexity-moderate, model-request
#803 - Circular dependency resolution
Pull Request -
State: closed - Opened by bryce13950 2 months ago
#802 - bumped python min version
Pull Request -
State: closed - Opened by bryce13950 2 months ago
#801 - [Bug Report] Padding side inconsistency with Huggingface Transformers
Issue -
State: open - Opened by spfrommer 2 months ago
Labels: bug, complexity-moderate, needs-investigation
#800 - [Bug Report] Load model problem
Issue -
State: open - Opened by LiuJinzhe-Keepgoing 2 months ago
- 4 comments
Labels: question
#799 - clarified arguments a bit for hook_points
Pull Request -
State: closed - Opened by bryce13950 2 months ago
#798 - [Proposal] Remove instances where `model_args` and `model_kwargs` are provided
Issue -
State: open - Opened by bryce13950 2 months ago
Labels: complexity-simple, breaking-change
#797 - Actions token access
Pull Request -
State: closed - Opened by bryce13950 2 months ago
#796 - [Proposal] change FactoredMatric.svd() so it doesn't prevent all instances of FactoredMatrix from being garbage collected
Issue -
State: open - Opened by manulari 3 months ago
- 1 comment
Labels: complexity-moderate
#795 - Raise exception when BERT is loaded with HookedTransformer instead of…
Pull Request -
State: closed - Opened by degenfabian 3 months ago
- 1 comment
#794 - Colab compatibility bug fixes
Pull Request -
State: closed - Opened by degenfabian 3 months ago
#793 - Remove einsum usage in forward function of BertMLMHead
Pull Request -
State: closed - Opened by degenfabian 3 months ago
#792 - Remove einsum usage in _get_w_in_matrix in SVDInterpreter
Pull Request -
State: closed - Opened by degenfabian 3 months ago
#791 - Remove einsum usage in refactor_factored_attn_matrices in HookedTransformer
Pull Request -
State: closed - Opened by degenfabian 3 months ago
#790 - Remove einsum in complex_attn_linear
Pull Request -
State: closed - Opened by degenfabian 3 months ago
- 2 comments
#789 - Remove einsum in compute_head_results in ActivationCache
Pull Request -
State: closed - Opened by degenfabian 3 months ago
#788 - Remove einsum in logit_attrs in ActivationCache
Pull Request -
State: closed - Opened by degenfabian 3 months ago
#787 - v2.9.1
Pull Request -
State: closed - Opened by bryce13950 3 months ago
#786 - added typeguard dependency
Pull Request -
State: closed - Opened by bryce13950 3 months ago
#785 - [Bug Report] Error on import due to missing `typeguard` dependency
Issue -
State: closed - Opened by chanind 3 months ago
- 1 comment
#784 - [Question] How to load a model in smaller precision?
Issue -
State: open - Opened by MittelmanDaniel 3 months ago
- 5 comments
Labels: needs-information
#783 - Remove einsum in forward pass in AbstractAttention
Pull Request -
State: closed - Opened by degenfabian 3 months ago
- 2 comments
#782 - Remove einsum in apply_causal_mask in abstract_attention.py
Pull Request -
State: closed - Opened by degenfabian 3 months ago
- 1 comment
#781 - Remove einsum usage from create_alibi_bias function
Pull Request -
State: closed - Opened by degenfabian 3 months ago
- 1 comment
#780 - Release 2.9
Pull Request -
State: closed - Opened by bryce13950 3 months ago
#779 - Add model upload and load
Pull Request -
State: open - Opened by mntss 3 months ago
#778 - [Bug Report] Global and Local Attn layer order of Gemma2 is wrong?
Issue -
State: open - Opened by huangxt39 3 months ago
Labels: complexity-moderate, implementation-inaccuracy
#777 - Fix that if use_past_kv_cache is set to True models from the Bloom family produce weird outputs.
Pull Request -
State: closed - Opened by degenfabian 3 months ago
#776 - [Bug Report] use_past_kv_cache yields weird outputs when used with Bloom model family
Issue -
State: closed - Opened by degenfabian 3 months ago
Labels: complexity-moderate
#775 - Set prepend_bos to false by default for Bloom model family
Pull Request -
State: closed - Opened by degenfabian 3 months ago
- 6 comments
#774 - [Proposal] prepend_bos should by default be set to false for the Bloom model family
Issue -
State: closed - Opened by degenfabian 3 months ago
Labels: complexity-moderate
#773 - [Question] Would it be possible to adopt TransformerLens on models with a different layernorm implementation?
Issue -
State: open - Opened by Steven-Yiran 3 months ago
- 2 comments
Labels: question, complexity-high
#772 - fix the bug that attention_mask and past_kv_cache cannot work together
Pull Request -
State: closed - Opened by yzhhr 3 months ago
#771 - Would you like to support models from the Qwen 2.5 series?
Issue -
State: closed - Opened by ArcherShirou 3 months ago
- 1 comment
#770 - Restore consistency of hook_normalized between LayerNorm and RMSNorm
Pull Request -
State: open - Opened by degenfabian 3 months ago
- 1 comment
#769 - improve model properties table in docs
Pull Request -
State: open - Opened by mivanit 3 months ago
- 6 comments
#768 - Add support for `Mistral-Nemo-Base-2407` model (#751)
Pull Request -
State: closed - Opened by frances720 3 months ago
#767 - v2.8.1
Pull Request -
State: closed - Opened by bryce13950 4 months ago
#766 - [Question] Question title
Issue -
State: closed - Opened by MittelmanDaniel 4 months ago
- 4 comments
#765 - Logit comparator tool
Pull Request -
State: closed - Opened by curt-tigges 4 months ago
#764 - Add support for NTK-by-Part Rotary Embedding & set correct rotary base for Llama-3.1 series
Pull Request -
State: closed - Opened by Hzfinfdu 4 months ago
- 2 comments
#763 - Add support for NTK-by-Part Rotary Embedding & set correct rotary base for Llama-3.1 series
Pull Request -
State: closed - Opened by Hzfinfdu 4 months ago
#762 - [Question] compatibility for 'Qwen/Qwen2.5-14B'
Issue -
State: closed - Opened by hgftrdw45ud67is8o89 4 months ago
- 1 comment
Labels: complexity-moderate, model-request
#761 - Add configurations for Llama 3.1 models(Llama-3.1-8B and Llama-3.1-70B)
Pull Request -
State: closed - Opened by vatsalrathod16 4 months ago
- 2 comments
#760 - Add py.typed for type hints
Pull Request -
State: open - Opened by UFO-101 4 months ago
- 2 comments
#759 - New issue template for reporting model compatibility
Pull Request -
State: closed - Opened by bryce13950 4 months ago
#758 - added new block for recent diagram, and colab compatibility notebook
Pull Request -
State: closed - Opened by bryce13950 4 months ago
#757 - Add warning and halt execution for incorrect T5 model usage
Pull Request -
State: closed - Opened by vatsalrathod16 4 months ago
#756 - Release 2.8
Pull Request -
State: closed - Opened by bryce13950 4 months ago