INDEX

Explanations

instances of the word "appeal."

oai_token-act-pair · gpt-3.5-turbo

references to appeals made to various groups or demographics

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

sterdam

-0.88

llan

-0.75

pite

-0.75

hesda

-0.72

alde

-0.65

lance

-0.64

 unused

-0.63

 Laur

-0.63

 awarded

-0.63

ikes

-0.61

POSITIVE LOGITS

 gauge

0.86

 caution

0.76

å§«

0.73

asty

0.70

strings

0.68

 reconsider

0.68

 reassure

0.68

Gau

0.67

 heights

0.66

 impulse

0.65

Activations Density 0.083%

instances of the word "appeal."

references to appeals made to various groups or demographics

No Comments

No Known Activations