AlignmentResearch/tuned-lens

Lamma-3-8b (#48)

Llama 2 7b (#45)

LLaMA-2-13b chat (#44)

LLaMA2-13b Lens (#43)

Revert "LLaMa2-7b tuned lens (#41)"

LLaMa2-7b tuned lens (#41)

switched back to 32 bit

Fixed plot title.

fixed several bugs in app.py

upgraded app to use tuned_lens=0.1.0

Remove broken unembeding hashes (#39)

Updated docker file

moved ln to earlier in the docker file

trying to fix dns issues

switched to docker

reset to using built in gradio applet

added .git to docker file

switched to using dockerfile

added docker ignore

fixed typo in name lmsys

llama-65b (#38)

vicuna (#37)

renamed lens

fixed lfs tracking

Vicuna (#36)

Add LLaMA models (#35)

Add pythia-2.8b-deduped-v0 (#34)

Added new pythia-deduped models (#33)

fixed lens version in so that app.py should continue to function

Reorginize lenses (#32)

Lens Migration (#29)

Remove broken lamma models (#28)

pythia-rename (#27)

Upload lens/llama-13b/config.json with huggingface_hub (#23)

Upload lens/llama-13b/params.pt with huggingface_hub (#24)

Upload lens/llama-7b/params.pt with huggingface_hub (#25)

Upload lens/llama-7b/config.json with huggingface_hub (#26)

Upload lens/gpt-neox-20b/params.pt with huggingface_hub (#21)

Upload lens/gpt-neox-20b/config.json with huggingface_hub (#22)

updated default text

updated examples

Added examples

removed dark theme from plot

App now runs by default and has better documentation

removed dark theme from plot

Changed to stride 2 and switched examine to submit

Upload lens/opt-125m/params.pt with huggingface_hub (#15)

Upload lens/opt-125m/config.json with huggingface_hub (#16)

Upload lens/opt-1.3b/params.pt with huggingface_hub (#17)

Upload lens/opt-1.3b/config.json with huggingface_hub (#18)