Inseq
@InseqLib
Open-Source Interpretability for Generative Language Models ๐ ๐
You might like
Hello world! ๐
After a year of restless development, I'm finally happy to announce Inseq, a new tool to democratize post-hoc interpretability of sequence generation models ๐ github.com/inseq-team/insโฆ #nlproc #xai Some highlights ๐ 1/
Model Internals-based RAG Evaluation (MIRAGE) ๐ด is accepted to #EMNLP2024 Main! โก๏ธ To celebrate, here's our new MIRAGE demo combining @InseqLib and Transformers-specific LRP: huggingface.co/spaces/gsarti/โฆ. Reach out if you want to catch up in Miami! ๐ค๐๏ธ
huggingface.co
MIRAGE - a Hugging Face Space by gsarti
MIRAGE - a Hugging Face Space by gsarti
[1/8] Struggling with verifying the trustworthiness of RAG outputs? Check our latest work where we utilize *model internals* as a powerful and faithful tool for attributing answers to retrieved docs! (w/ @gsarti_ @AriannaBisazza @raquel_dmg) ๐: arxiv.org/abs/2406.13663 #NLProc
![Jirui_Qi's tweet image. [1/8] Struggling with verifying the trustworthiness of RAG outputs? Check our latest work where we utilize *model internals* as a powerful and faithful tool for attributing answers to retrieved docs! (w/ @gsarti_ @AriannaBisazza @raquel_dmg)
๐: arxiv.org/abs/2406.13663
#NLProc](https://pbs.twimg.com/media/GQnSi2wXsAADZ2Z.jpg)
Very hyped for the new beautiful viz that just landed in the @InseqLib main branch! ๐ฅ This will empower users to explore attribution tensors more flexibly and intuitively. h/t to @_ddjohnson for his awesome work on the treescope toolkit powering this release!
Thanks to the new treescope integration, @InseqLib now supports interactive visualizations for multidimensional attributions (show_granular), token highlights (show_tokens) and improved viz for attribute_context CLI! ๐ Install main, will appear in v0.7 x.com/_ddjohnson/staโฆ
Thanks to the new treescope integration, @InseqLib now supports interactive visualizations for multidimensional attributions (show_granular), token highlights (show_tokens) and improved viz for attribute_context CLI! ๐ Install main, will appear in v0.7 x.com/_ddjohnson/staโฆ
By popular demand, the Treescope pretty-printer from the Penzai neural net library can now be installed separately, and supports both JAX and PyTorch! And that's not all: Penzai itself now has less boilerplate and includes more pretrained Transformer models!
๐ We're thrilled to host Gabriele Sarti (@gsarti_) in our PhD seminar series tomorrow, July 16th, from 12:00-13:00 in Oe67 BU 101! Join us for his talk on interpreting context usage in generative language models, featuring the Inseq toolkit and PECoRe framework. ๐Don't miss it!
The ๐ PECoRe / ๐ด MIRAGE demo on @huggingface Spaces is powered by our new attribute-context CLI command released in v0.6, and allows to export the code to reproduce your results locally with ๐ Inseq. Check it out โก๏ธ hf.co/spaces/gsarti/โฆ
โ ๏ธ Citations from prompting or NLI seem plausible, but may not faithfully reflect LLM reasoning. ๐๏ธ MIRAGE detects context dependence in generations via model internals, producing granular and faithful RAG citations. ๐ Demo: huggingface.co/spaces/gsarti/โฆ Fun collab w/ @Jirui_Qi,โฆ
Today, we had the first seminar of our #XAI course! @gsarti_ presented the @InseqLib to interpret LMs and the PECORE framework to identify & attribute context dependence in LMs! ๐๐ Thank you, it was so interesting! ๐ค Great start to our series! gsarti.com/talk/polito-inโฆ
[1/4] Introducing โA Primer on the Inner Workings of Transformer-based Language Modelsโ, a comprehensive survey on interpretability methods and the findings into the functioning of language models they have led to. ArXiv: arxiv.org/pdf/2405.00208
![javifer_96's tweet image. [1/4] Introducing โA Primer on the Inner Workings of Transformer-based Language Modelsโ, a comprehensive survey on interpretability methods and the findings into the functioning of language models they have led to.
ArXiv: arxiv.org/pdf/2405.00208](https://pbs.twimg.com/media/GMpDM85WcAELzoF.jpg)
Today @InseqLib hit 300 โญ๏ธ on Github! A huge thank you to all our awesome users โค๏ธ Onwards to the next 300! ๐คบ
@InseqLib v0.6 is out now on PyPI! ๐ฅ New CLI command for context attribution (@gsarti_), new perturbation-based methods by @hmohebbi75 & @casszzx and optimizations incl. multi-gpu support! โก๏ธ Huge shoutout to our contributors! โค๏ธ Release notes โฌ๏ธ github.com/inseq-team/insโฆ
Want to learn how to detect and attribute context usage in LMs using @InseqLib? This @Gradio demo using our CLI can teach you how! ๐
The official ๐ PECoRe ๐ demo to detect and attribute context dependence in LM generations is now available on @huggingface Spaces! ๐ Includes code examples, a usage guide, useful presets for various dec-only & enc-dec models, and more! Check it out โฌ๏ธ huggingface.co/spaces/gsarti/โฆ
๐ PECoRe repository is now public (github.com/gsarti/pecore) and all model/datasets are available on @huggingface (huggingface.co/collections/gsโฆ)! ๐Interested in using PECoRe on your models? Have a look at the @InseqLib implementation (`inseq attribute-context`)!
huggingface.co
๐๐ PECoRe @ ICLR 2024 - a gsarti Collection
๐๐ PECoRe @ ICLR 2024 - a gsarti Collection
[1/8] Our new work (w/ @AriannaBisazza @gchrupala @MalvinaNissim) is finally out! ๐ We introduce PECoRe, an interpretability framework using model internals to identify & attribute context dependence in language models. ๐Paper: arxiv.org/abs/2310.01188 #NLProc #neuralempty
![gsarti_'s tweet image. [1/8] Our new work (w/ @AriannaBisazza @gchrupala @MalvinaNissim) is finally out! ๐
We introduce PECoRe, an interpretability framework using model internals to identify & attribute context dependence in language models.
๐Paper: arxiv.org/abs/2310.01188
#NLProc #neuralempty](https://pbs.twimg.com/media/F7lQGhnWsAEfWCX.jpg)
Excited to present my recent work on @InseqLib and the #ICLR2024 PECoRe interpretability framework for the @SheffieldNLP group this afternoon! Many thanks @casszzx for inviting me! ๐ค ๐ Inseq: github.com/inseq-team/insโฆ ๐๐ PECoRe: openreview.net/forum?id=XTHfNโฆ
Wow!! This is great news๐ฅณ Thanks @InseqLib! Iโm excited to try it out!
Value Zeroing, a faithful approach for analyzing context mixing in Transformers, is now available on @InseqLib main branch for all @huggingface text generation models! ๐ ๐Paper introducing VZ: aclanthology.org/2023.eacl-mainโฆ ๐VZ in Inseq: tinyurl.com/inseq-vz
After two years with Poetry, Inseq just moved to @astral_sh's new blazing fast package manager uv! Our CI installation step is now ~80% faster! ๐ฅ Congrats to @charliermarsh and the team on the release, and godspeed for the cargo-like experience you are planning to build!
United States Trends
- 1. #SpotifyWrapped 40.8K posts
- 2. Chris Paul 17.2K posts
- 3. Clippers 26.3K posts
- 4. #NSD26 7,784 posts
- 5. Good Wednesday 30.1K posts
- 6. #WednesdayMotivation 3,501 posts
- 7. Hump Day 10.8K posts
- 8. National Signing Day 3,459 posts
- 9. Happy Hump 7,141 posts
- 10. Nashville 31.9K posts
- 11. TOP CALL 10.6K posts
- 12. #Wednesdayvibe 1,938 posts
- 13. FELIX LV VISIONARY SEOUL 28.2K posts
- 14. The BIGGะST 309K posts
- 15. #ALLOCATION 258K posts
- 16. Wordle 1,628 X N/A
- 17. Welcome Home 14.4K posts
- 18. Somalis 103K posts
- 19. Fisherman 11K posts
- 20. Elden Campbell 1,059 posts
You might like
-
Gabriele Sarti
@gsarti_ -
Arianna Bisazza #EMNLP ๐จ๐ณ
@AriannaBisazza -
AILC_NLP
@AILC_NLP -
BeatriceSavoldi
@BeatriceSavoldi -
EVALITA
@EVALITAcampaign -
Ahmet รstรผn @NeurIPS โ๏ธ
@ahmetustun89 -
Javier Ferrando
@javifer_96 -
Michael Hanna@NeurIPS2025
@michaelwhanna -
Marco Gaido
@mgaido91 -
Hosein Mohebbi
@hmohebbi75 -
Nils Feldhus
@nfelnlp -
Alessio Miaschi
@AlessioMiaschi -
Huiyuan Lai
@HuiyuanLai -
Marco Guerini
@m_guerini -
Lukas Edman
@LukasNLP
Something went wrong.
Something went wrong.