INDEX

Explanations

complex words that are related to completion or enhancement

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 laughter

-0.71

 Baghd

-0.64

 spine

-0.63

 Sabb

-0.59

ISO

-0.59

 Franch

-0.58

ocker

-0.58

bucks

-0.58

rooms

-0.57

isms

-0.57

POSITIVE LOGITS

mented

1.75

teness

1.44

ased

1.32

ments

1.30

mental

1.13

mentation

1.13

ment

1.12

asing

1.09

MENT

1.06

ĵ

1.01

Activations Density 0.050%

complex words that are related to completion or enhancement

No Comments

No Known Activations