INDEX

Explanations

contractions ending with 't

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "can't" or variations that express inability or negation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

aldi

-0.85

ranged

-0.70

pires

-0.69

furt

-0.69

emetery

-0.66

oak

-0.65

imity

-0.65

iferation

-0.64

 arsen

-0.64

pocket

-0.64

POSITIVE LOGITS

 imagine

1.09

 emphasize

1.08

 stress

0.97

 remember

0.95

 attest

0.95

athom

0.94

 believe

0.94

 recommend

0.93

 thank

0.93

 conceive

0.93

Activations Density 0.049%

contractions ending with 't

instances of the phrase "can't" or variations that express inability or negation

No Comments

No Known Activations