INDEX

Explanations

codes or identifiers in a particular format

oai_token-act-pair · gpt-3.5-turbo

references to government agencies or organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 socialism

-0.71

 communism

-0.71

 Reck

-0.70

 spont

-0.63

 decad

-0.61

 companions

-0.60

 adventurous

-0.60

opp

-0.59

 Czech

-0.59

 fiance

-0.58

POSITIVE LOGITS

GA

4.48

GA

1.95

ESA

1.61

ga

1.54

Ga

1.54

GS

1.47

GI

1.42

GG

1.38

GE

1.36

GF

1.34

Activations Density 0.014%

codes or identifiers in a particular format

references to government agencies or organizations

No Comments

No Known Activations