INDEX

Explanations

mentions of medical care or attention to health-related matters

oai_token-act-pair · gpt-3.5-turbo

references to various types of care

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

ãĥĥãĥī

-0.81

kered

-0.80

ãĥ³ãĤ¸

-0.79

onite

-0.68

 bluff

-0.67

âĸ¬

-0.65

akedown

-0.63

 repr

-0.62

awe

-0.61

 Mines

-0.61

POSITIVE LOGITS

taker

1.53

giving

1.23

lessly

1.07

tta

1.04

taking

1.03

fully

0.93

lessness

0.91

lington

0.88

free

0.86

maid

0.86

Activations Density 0.025%

mentions of medical care or attention to health-related matters

references to various types of care

No Comments

No Known Activations