INDEX

Explanations

phrases indicating errors or issues

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "wrong."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

otine

-0.85

urst

-0.82

ilt

-0.82

gil

-0.77

ocry

-0.76

thin

-0.76

cit

-0.75

 brim

-0.73

ilet

-0.72

prise

-0.71

POSITIVE LOGITS

 Mara

0.62

 Malaysian

0.61

ario

0.60

cies

0.60

 Investigator

0.60

 attendant

0.59

Pav

0.59

 plag

0.58

akia

0.58

 Manor

0.58

Activations Density 0.041%

phrases indicating errors or issues

occurrences of the word "wrong."

No Comments

No Known Activations