INDEX

Explanations

document and news-related metadata, such as titles, subtitles, and sources

oai_token-act-pair · gpt-3.5-turbo

sentences that reveal significant facts or events

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

 paddle

-0.97

 vegetarian

-0.95

 regener

-0.89

 environmentally

-0.85

 healthier

-0.85

 carniv

-0.85

 rebirth

-0.85

 sustainable

-0.84

 sustainability

-0.84

 happiness

-0.83

POSITIVE LOGITS

Documents

1.71

Investigators

1.68

FBI

1.61

CIA

1.58

 Investigators

1.53

Wiki

1.48

Leaks

1.45

 Documents

1.45

NSA

1.44

Prosecutors

1.40

Activations Density 0.496%

document and news-related metadata, such as titles, subtitles, and sources

sentences that reveal significant facts or events

No Comments

No Known Activations