INDEX

Explanations

letter sequences that are specific to a particular encoding format

oai_token-act-pair · gpt-3.5-turbo

references to significant literary works and their connections or representations

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

 wont

-0.87

 tremend

-0.83

ettings

-0.77

 neigh

-0.75

 fuzz

-0.70

chool

-0.69

chini

-0.69

ansky

-0.68

ecided

-0.67

 ende

-0.67

POSITIVE LOGITS

Trivia

0.88

Appearances

0.80

Notable

0.79

Quotes

0.77

Gallery

0.76

References

0.73

 Alternatively

0.72

Contents

0.71

See

0.69

History

0.68

Activations Density 0.209%

letter sequences that are specific to a particular encoding format

references to significant literary works and their connections or representations

No Comments

No Known Activations