INDEX

Explanations

references to news headlines and public statements

oai_token-act-pair · gpt-3.5-turbo

significant events and their consequences

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

NetMessage

-0.78

ĵĺ

-0.64

ĺħ

-0.56

ADRA

-0.51

rencies

-0.48

 fixme

-0.47

azines

-0.47

*/(

-0.47

theless

-0.46

 Converted

-0.46

POSITIVE LOGITS

icter

0.69

ieth

0.63

iest

0.59

iets

0.54

liest

0.54

 slate

0.52

 portion

0.52

Versions

0.51

liner

0.51

hest

0.51

Activations Density 7.524%

references to news headlines and public statements

significant events and their consequences

No Comments

No Known Activations