INDEX

Explanations

mentions of working on various projects

oai_token-act-pair · gpt-3.5-turbo

instances of the word "worked."

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

ylon

-0.67

arta

-0.66

gran

-0.64

Bol

-0.63

idium

-0.63

ustomed

-0.63

Mae

-0.62

thur

-0.62

 Tian

-0.62

 venerable

-0.61

POSITIVE LOGITS

bench

0.94

hops

0.93

 worked

0.88

hirt

0.88

 ethic

0.86

 collabor

0.85

heet

0.79

 overtime

0.76

 arrang

0.75

baugh

0.74

Activations Density 0.021%

mentions of working on various projects

instances of the word "worked."

No Comments

No Known Activations