INDEX

Explanations

photographs or photo-related references

oai_token-act-pair · gpt-3.5-turbo

mentions of "Photo" or references to images

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

amped

-0.69

 footing

-0.68

 foreseeable

-0.63

 judgement

-0.59

amping

-0.59

iated

-0.59

oes

-0.57

outine

-0.54

oret

-0.54

 impending

-0.54

POSITIVE LOGITS

journal

1.12

 illustration

1.02

 Gallery

0.98

 courtesy

0.96

 Credit

0.96

 gallery

0.93

 Illust

0.91

 Courtesy

0.89

 credit

0.88

0.87

Activations Density 0.022%

photographs or photo-related references

mentions of "Photo" or references to images

No Comments

No Known Activations