INDEX

Explanations

words related to specific names such as "Dust", "Gust", and "Huston"

oai_token-act-pair · gpt-3.5-turbo

references to the term "Dust."

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

 unpre

-0.76

————

-0.71

thood

-0.65

iannopoulos

-0.64

bably

-0.63

isin

-0.63

--------------------------------------------------------

-0.62

perty

-0.61

cers

-0.60

ntil

-0.60

POSITIVE LOGITS

 Dust

1.23

Dust

1.04

dust

0.92

meyer

0.87

schild

0.80

 Tornado

0.80

 Rating

0.76

works

0.74

forge

0.74

 Crate

0.73

Activations Density 0.010%

words related to specific names such as "Dust", "Gust", and "Huston"

references to the term "Dust."

No Comments

No Known Activations