INDEX

Explanations

words related to criticism or disagreement

oai_token-act-pair · gpt-3.5-turbo

phrases related to being criticized or deemed inappropriate

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ositories

-0.68

iosyncr

-0.64

Jump

-0.64

Query

-0.63

Impro

-0.63

Roberts

-0.62

 strand

-0.62

orgetown

-0.62

rub

-0.61

dfx

-0.61

POSITIVE LOGITS

alls

0.88

ength

0.85

ogue

0.84

arding

0.83

arded

0.83

worth

0.82

ares

0.78

alled

0.77

BuyableInstoreAndOnline

0.77

Activations Density 0.009%

words related to criticism or disagreement

phrases related to being criticized or deemed inappropriate

No Comments

No Known Activations