INDEX

Explanations

references to concealed or obscured information

oai_token-act-pair · gpt-3.5-turbo

references to concepts or ideas that are concealed or not immediately obvious

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

annis

-0.86

nia

-0.79

sburgh

-0.79

alon

-0.78

Pwr

-0.76

night

-0.74

ctive

-0.73

umbn

-0.73

rehens

-0.73

Ħ¢

-0.71

POSITIVE LOGITS

gem

0.95

 recess

0.86

 gems

0.86

 hidden

0.84

 behind

0.79

 Hidden

0.79

 treasures

0.76

 underground

0.75

 ingredient

0.74

 hatch

0.73

Activations Density 0.021%

references to concealed or obscured information

references to concepts or ideas that are concealed or not immediately obvious

No Comments

No Known Activations