INDEX

Explanations

phrases related to editing and drafting

oai_token-act-pair · gpt-3.5-turbo

references to edits or changes in content

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

 PLUS

-0.76

 neighb

-0.73

hement

-0.69

 Learns

-0.66

 Oops

-0.64

 surpr

-0.64

 servicing

-0.62

BIG

-0.61

 lifes

-0.61

DK

-0.60

POSITIVE LOGITS

References

0.99

See

0.88

tymology

0.86

Trivia

0.85

Flag

0.82

Quotes

0.81

Appearances

0.81

Gallery

0.81

External

0.79

Introduced

0.79

Activations Density 0.112%

phrases related to editing and drafting

references to edits or changes in content

No Comments

No Known Activations