INDEX

Explanations

references to the concept of 'witch hunts'

oai_token-act-pair · gpt-3.5-turbo

references to witch hunts and witch-related terminology

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

phia

-0.77

 IMAGES

-0.72

upon

-0.68

ais

-0.67

interrupted

-0.65

inished

-0.63

 Fault

-0.63

IGH

-0.62

andro

-0.62

ateral

-0.61

POSITIVE LOGITS

doctor

1.22

hunt

0.93

sonian

0.91

haz

0.89

ry

0.89

ery

0.86

finder

0.83

tail

0.81

hattan

0.81

 Hazel

0.79

Activations Density 0.049%

references to the concept of 'witch hunts'

references to witch hunts and witch-related terminology

No Comments

No Known Activations