INDEX

Explanations

visual descriptions or representations

oai_token-act-pair · gpt-3.5-turbo

mentions of imagery and visuals in a document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Buff

-0.72

ente

-0.70

acs

-0.69

pc

-0.68

ichi

-0.63

cc

-0.63

grad

-0.63

 broke

-0.63

ambers

-0.61

idel

-0.61

POSITIVE LOGITS

 imagery

3.56

 symbolism

1.76

 visuals

1.58

 Imag

1.32

 metaphors

1.30

 images

1.26

 artwork

1.20

 visualization

1.19

 motif

1.12

 footage

1.12

Activations Density 0.018%

visual descriptions or representations

mentions of imagery and visuals in a document

No Comments

No Known Activations