INDEX

Explanations

references or mentions of specific formats or formats themselves

oai_token-act-pair · gpt-3.5-turbo

references to various types of formats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

doms

-0.93

roma

-0.88

adows

-0.76

arma

-0.72

minent

-0.69

atana

-0.68

yer

-0.68

ghan

-0.68

riv

-0.68

nee

-0.67

POSITIVE LOGITS

 format

0.98

ters

0.91

 formats

0.87

 Format

0.86

ftime

0.82

atted

0.79

furt

0.74

 formatted

0.74

Feature

0.73

ting

0.72

Activations Density 0.022%

references or mentions of specific formats or formats themselves

references to various types of formats

No Comments

No Known Activations