issues
search
TransformerLensOrg
/
TransformerLens
A library for mechanistic interpretability of GPT-style language models
https://transformerlensorg.github.io/TransformerLens/
MIT License
1.45k
stars
283
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
[Proposal] Add Lllama 3.1 support
#691
ssuukk
opened
2 months ago
14
Python 3.8 removal
#690
bryce13950
closed
2 months ago
0
Added gemma-2 2b (#687)
#689
bryce13950
closed
2 months ago
0
2.3.0
#688
bryce13950
closed
2 months ago
0
Added gemma-2 2b
#687
curt-tigges
closed
2 months ago
1
OSError: gpt2 does not appear to have a file named config.json. Checkout 'https://huggingface.co/gpt2/None' for available files.
#686
Iust1n2
closed
2 days ago
0
[Bug Report] Different results from HuggingFace when using the GPT2 small example
#685
nreHieW
opened
2 months ago
0
[Question] Why does Transformer Lens only support quantized LLaMA models?
#684
miguel-kjh
opened
2 months ago
1
[Bug Report] Qwen model implementation is too inaccurate
#683
bryce13950
opened
2 months ago
3
updated dependencies
#682
bryce13950
opened
2 months ago
0
Test arena cleanup
#681
bryce13950
closed
2 months ago
0
[Proposal] Demo and Tutorial on Patchscopes and "Patching + Generation"
#680
HenryCai11
closed
1 month ago
5
NamesFilter can be a string
#679
jettjaniak
closed
2 months ago
1
Add Mixtral to `test_match_huggingface` test
#678
joelburget
closed
3 weeks ago
1
Fix typo in `embed.py` docs
#677
ArthurConmy
closed
2 months ago
0
Move the HookedSAE / HookedSAETransformer warning to a less prominent…
#676
ArthurConmy
closed
2 months ago
0
Release 2.2.2
#675
bryce13950
closed
2 months ago
0
added arena content as a notebook
#674
bryce13950
closed
2 months ago
0
fix: fixing broken backward hooks change
#673
chanind
closed
2 months ago
1
[Bug Report] Backward hooks are broken as of v2.0.0
#672
chanind
closed
2 months ago
1
[Proposal] Allow tied embeddings
#671
neelnanda-io
opened
2 months ago
1
ValueError: microsoft/Phi-3-mini-128k-instruct not found.
#670
joykirat18
opened
2 months ago
1
does run_with_cache method support data parallel , how can I do it ?
#669
Yang-bug-star
opened
2 months ago
0
Release 2.2.1
#668
bryce13950
closed
2 months ago
0
[Bug Report] Einops shape error when `use_attn_result = True`
#667
dtch1997
closed
2 months ago
1
Fix attention result projection
#666
callummcdougall
closed
2 months ago
2
[Proposal] Allow recent versions of beartype
#665
jettjaniak
opened
2 months ago
6
[Question] Offline Error HookedTransformer.from_pretrained
#664
pbernabeup
closed
2 months ago
3
Adding RMSNorm to apply_ln_to_stack
#663
gaabrielfranco
closed
2 months ago
1
Add support for Qwen2 models
#662
g-w1
closed
2 months ago
3
[Bug Report] Pythia output inconsistent across batch sizes when use_split_qkv_input=True
#661
oliveradk
opened
2 months ago
0
removed einsum causing error when use_atten_result is enabled
#660
oliveradk
closed
1 month ago
2
[Bug Report] Attn Result hook not working
#659
oliveradk
closed
2 months ago
2
docs: update Main_Demo.ipynb
#658
eltociear
closed
2 months ago
1
[Bug Report] RMSNormPre in Transformer_lens is maybe different from Llama source code?
#657
wangyifei0047
opened
2 months ago
1
Release 2.2
#656
bryce13950
closed
2 months ago
0
Is it possible to use a locally downloaded model without accessing HF?
#655
ccp123456
opened
3 months ago
9
Fix Out bias not being summed in attention component when using 4 bit precision
#654
FlyingPumba
closed
3 months ago
1
[Question] loading Llama3-8B-instruct to HookedTransformer got a warning saying You are not using LayerNorm, so the writing weights can't be centered! Skipping!
#653
wangyifei0047
closed
3 months ago
1
Mlp cleanup
#652
bryce13950
closed
2 months ago
0
[Bug Report] Phi-3 Model does not load on Transformer Lens
#651
KanishkT123
closed
3 months ago
3
Added support for Gemma-2
#650
neelnanda-io
closed
2 months ago
11
Model baichuan
#649
bryce13950
opened
3 months ago
0
Fixed weight conversion
#648
bryce13950
closed
3 months ago
0
Move out pretrained weight conversions
#647
richardkronick
closed
3 months ago
0
Moved mixtral weights to another module
#646
bryce13950
closed
3 months ago
0
Match Huggingface GPT2 implementation *exactly*
#645
joelburget
closed
3 months ago
2
[Proposal] Documentation: Map the Act Names to the Transformer
#644
JuVogt
opened
3 months ago
3
Add tests for ActivationCache
#643
FlyingPumba
closed
3 months ago
5
Steering vanilla GPT2 with SAE vectors based on transformerlens version of GPT2
#642
ianand
closed
3 months ago
3
Previous
Next