INDEX

Explanations

comparisons where one thing is significantly outweighing another

oai_token-act-pair · gpt-3.5-turbo

concepts related to comparison and evaluation of effects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

lear

-0.85

 whore

-0.63

aire

-0.63

olog

-0.63

natureconservancy

-0.62

 charism

-0.62

RH

-0.61

opath

-0.61

ionics

-0.61

eg

-0.59

POSITIVE LOGITS

ighed

1.22

 outweigh

1.00

 outwe

0.86

ĸļ

0.81

¿½

0.81

ĺħ

0.80

INGTON

0.79

swer

0.78

olulu

0.76

edIn

0.74

Activations Density 0.016%

comparisons where one thing is significantly outweighing another

concepts related to comparison and evaluation of effects

No Comments

No Known Activations