INDEX

Explanations

proper nouns, likely related to news titles or headlines

oai_token-act-pair · gpt-3.5-turbo

references to ecological or environmental topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Mysterious

-0.79

Sul

-0.72

 wings

-0.71

 Hanson

-0.71

sab

-0.66

 Missouri

-0.66

-0.65

 siph

-0.65

 Indra

-0.65

 fallout

-0.65

POSITIVE LOGITS

ot

1.31

ata

1.10

oton

1.06

at

1.05

ato

1.02

AT

1.01

atum

0.98

vet

0.97

atell

0.97

et

0.97

Activations Density 0.262%

proper nouns, likely related to news titles or headlines

references to ecological or environmental topics

No Comments

No Known Activations