INDEX

Explanations

words related to negative or problematic situations that are causing concern or difficulty

oai_token-act-pair · gpt-3.5-turbo

instances of the word "trouble."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

oliberal

-0.85

ulture

-0.80

uine

-0.73

obook

-0.72

ithe

-0.72

itation

-0.71

inen

-0.70

ebook

-0.70

Interstitial

-0.70

tein

-0.68

POSITIVE LOGITS

hooting

0.97

makers

0.85

 troubles

0.85

 trouble

0.84

maker

0.73

 interfering

0.72

 ..........

0.71

fully

0.71

 plag

0.70

lessly

0.70

Activations Density 0.016%

words related to negative or problematic situations that are causing concern or difficulty

instances of the word "trouble."

No Comments

No Known Activations