INDEX

Explanations

words related to seriousness or severity

oai_token-act-pair · gpt-3.5-turbo

mentions of serious topics or issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

enaries

-0.91

ifully

-0.82

 seamlessly

-0.81

wright

-0.78

rious

-0.74

anic

-0.73

sylv

-0.72

anol

-0.71

ilant

-0.71

atu

-0.70

POSITIVE LOGITS

 consideration

1.03

 contender

0.95

 doubts

0.94

 contenders

0.91

 repercussions

0.89

 injury

0.88

 threat

0.87

 jeopardy

0.86

 consequences

0.85

 threats

0.85

Activations Density 0.065%

words related to seriousness or severity

mentions of serious topics or issues

No Comments

No Known Activations