INDEX

Explanations

medical-related abbreviations or acronyms

oai_token-act-pair · gpt-3.5-turbo

references to specific statistical values or team-related metrics in sports contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

velt

-0.86

amus

-0.85

frames

-0.83

frame

-0.73

fram

-0.72

kamp

-0.71

sexual

-0.70

phen

-0.68

erville

-0.66

âĹ¼

-0.65

POSITIVE LOGITS

mpeg

1.22

ORTS

0.92

iants

0.90

ORD

0.89

IX

0.87

ranch

0.83

antasy

0.82

VII

0.81

OUND

0.80

ledged

0.80

Activations Density 0.008%

medical-related abbreviations or acronyms

references to specific statistical values or team-related metrics in sports contexts

No Comments

No Known Activations