INDEX

Explanations

dates or events described in chronological order

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, specifically semicolons

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 unnecess

-0.77

agic

-0.76

 millenn

-0.73

stration

-0.70

din

-0.68

ront

-0.67

orts

-0.66

 attacker

-0.66

agara

-0.66

 tremend

-0.65

POSITIVE LOGITS

-)

1.03

alias

0.86

 alternatively

0.83

âĢ¢âĢ¢âĢ¢âĢ¢

0.79

;;;;;;;;;;;;

0.76

cf

0.73

âĢ¢âĢ¢

0.70

 Editing

0.70

};

0.69

 thence

0.65

Activations Density 0.046%

dates or events described in chronological order

punctuation marks, specifically semicolons

No Comments

No Known Activations