INDEX

Explanations

instances where the phrase "don't" is used

oai_token-act-pair · gpt-3.5-turbo

expressions of uncertainty or refusal

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

ategory

-0.75

 afore

-0.73

artney

-0.72

ANG

-0.69

 Agency

-0.66

 Passage

-0.62

 Anim

-0.62

upp

-0.59

 Personality

-0.58

 Antar

-0.58

POSITIVE LOGITS

't

1.31

ned

0.89

uts

0.83

ates

0.79

ÃŃ

0.78

anted

0.75

nas

0.71

kie

0.70

na

0.69

itzer

0.69

Activations Density 0.075%

instances where the phrase "don't" is used

expressions of uncertainty or refusal

No Comments

No Known Activations