INDEX

Explanations

specific formatting patterns, such as unusual characters or sequences

oai_token-act-pair · gpt-3.5-turbo

punctuation and formatting elements in text

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 equival

-0.86

 reflex

-0.85

 instinct

-0.84

 shorthand

-0.84

 agon

-0.80

 persuasion

-0.79

 sucker

-0.78

pse

-0.78

 grip

-0.76

 revol

-0.75

POSITIVE LOGITS

Meanwhile

1.52

RELATED

1.49

Tickets

1.46

Also

1.46

Advertisements

1.45

Additionally

1.44

According

1.44

Among

1.43

About

1.42

Other

1.40

Activations Density 0.501%

specific formatting patterns, such as unusual characters or sequences

punctuation and formatting elements in text

No Comments

No Known Activations