INDEX

Explanations

strings of characters that seem to be special tokens or delimiters, potentially used for formatting or encoding purposes

oai_token-act-pair · gpt-3.5-turbo

incidents involving violence or crime

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 favour

-0.85

 renegoti

-0.83

 scept

-0.80

 monet

-0.78

 domestically

-0.77

 censor

-0.76

utsche

-0.76

 endeavour

-0.75

 endeav

-0.73

 moder

-0.73

POSITIVE LOGITS

RELATED

1.13

ccording

1.11

Police

1.10

About

1.08

Authorities

1.07

Contribut

1.05

Copyright

1.01

Investigators

1.01

SPONSORED

0.98

Officers

0.98

Activations Density 0.869%

strings of characters that seem to be special tokens or delimiters, potentially used for formatting or encoding purposes

incidents involving violence or crime

No Comments

No Known Activations