INDEX

Explanations

numbers and technical details, potentially related to transactions or specifications

oai_token-act-pair · gpt-3.5-turbo

numerical values, specifically those related to measurements or identifiers

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

alde

-0.97

atis

-0.86

atism

-0.82

icans

-0.78

omon

-0.77

ushi

-0.74

giving

-0.73

anamo

-0.72

ebin

-0.72

oken

-0.72

POSITIVE LOGITS

lihood

0.83

issance

0.73

à¼

0.64

weed

0.63

weight

0.63

æ©Ł

0.62

 needles

0.60

 plum

0.59

escription

0.59

yne

0.59

Activations Density 0.037%

numbers and technical details, potentially related to transactions or specifications

numerical values, specifically those related to measurements or identifiers

No Comments

No Known Activations