INDEX

Explanations

terms related to the concept of "witch hunt."

oai_token-act-pair · gpt-3.5-turbo

references to witch hunts or witches in various contexts

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

ais

-0.77

inished

-0.76

oulos

-0.69

undo

-0.69

icts

-0.68

ateral

-0.67

rez

-0.63

ouf

-0.63

ettel

-0.63

jri

-0.62

POSITIVE LOGITS

doctor

1.10

haz

1.05

 hunts

0.93

 witch

0.92

hunt

0.89

 Witch

0.88

witch

0.85

crow

0.84

 Hazel

0.83

finder

0.81

Activations Density 0.019%

terms related to the concept of "witch hunt."

references to witch hunts or witches in various contexts

No Comments

No Known Activations