INDEX

Explanations

words related to new paragraphs or breaks in text

oai_token-act-pair · gpt-3.5-turbo

sequences of punctuation or formatting used in unconventional ways

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 footing

-0.76

 antit

-0.76

hement

-0.71

 intended

-0.69

 estranged

-0.69

 extrad

-0.67

ulkan

-0.67

 furnished

-0.66

 ideal

-0.65

iton

-0.65

POSITIVE LOGITS

Anyway

1.46

 Anyway

1.12

ÂłÂłÂłÂł

1.08

Posted

1.07

ÂłÂłÂł

1.07

ÂłÂłÂłÂłÂłÂłÂłÂł

1.06

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

1.05

Seriously

1.04

Anonymous

1.04

Okay

0.99

Activations Density 0.396%

words related to new paragraphs or breaks in text

sequences of punctuation or formatting used in unconventional ways

No Comments

No Known Activations