INDEX

Explanations

phrases that suggest or imply something without explicitly stating it

oai_token-act-pair · gpt-3.5-turbo

terms that indicate causality or suggest a relationship between situations and their consequences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

rawdownloadcloneembedreportprint

-0.72

EEP

-0.69

 Maher

-0.68

iets

-0.68

foreseen

-0.68

hari

-0.66

annon

-0.65

 bottleneck

-0.65

Published

-0.64

thodox

-0.63

POSITIVE LOGITS

 indicative

1.03

 suggestive

1.00

ographically

0.92

notations

0.85

 chronological

0.83

ographical

0.80

 salv

0.75

ographic

0.71

ģ«

0.71

oral

0.71

Activations Density 0.020%

phrases that suggest or imply something without explicitly stating it

terms that indicate causality or suggest a relationship between situations and their consequences

No Comments

No Known Activations