INDEX

Explanations

documents with mentions of specific individuals and medical conditions

oai_token-act-pair · gpt-3.5-turbo

instances of reported speech or statements made by individuals

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

hub

-0.71

selves

-0.59

eps

-0.57

 Their

-0.55

atters

-0.52

 THEIR

-0.51

 unison

-0.51

odes

-0.51

 Hole

-0.51

 Recommended

-0.50

POSITIVE LOGITS

 himself

0.95

 remorse

0.74

 confess

0.69

 confessed

0.68

sul

0.66

 inco

0.65

 briefly

0.63

 calmly

0.63

 voluntarily

0.62

¶æ

0.62

Activations Density 1.481%

documents with mentions of specific individuals and medical conditions

instances of reported speech or statements made by individuals

No Comments

No Known Activations