INDEX

Explanations

proper nouns or names of people or places

oai_token-act-pair · gpt-3.5-turbo

phrases indicating widespread attention and scrutiny

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

à¦

-0.73

hemer

-0.60

odynam

-0.60

chell

-0.60

adelphia

-0.59

uum

-0.59

Ì¶

-0.59

ope

-0.59

oreal

-0.58

sectional

-0.58

POSITIVE LOGITS

 Fury

0.65

 Opposition

0.63

 Fowler

0.62

 Swanson

0.62

 Welch

0.59

 Stanton

0.56

 Spur

0.54

rium

0.53

 Honduras

0.52

 academia

0.51

Activations Density 1.258%

proper nouns or names of people or places

phrases indicating widespread attention and scrutiny

No Comments

No Known Activations