INDEX

Explanations

mentions of legal or criminal activities

oai_token-act-pair · gpt-3.5-turbo

instances of commas in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

swick

-0.70

herent

-0.61

 seams

-0.60

 doses

-0.59

olves

-0.57

OTUS

-0.57

 fleeting

-0.57

 combust

-0.57

agate

-0.56

aval

-0.55

POSITIVE LOGITS

 citing

0.89

 although

0.80

 quoting

0.80

 adding

0.78

 "âĢ¦

0.76

 noting

0.74

 albeit

0.73

"'

0.73

"(

0.70

 anecd

0.70

Activations Density 0.162%

mentions of legal or criminal activities

instances of commas in the text

No Comments

No Known Activations