INDEX

Explanations

mentions of hats

oai_token-act-pair · gpt-3.5-turbo

mentions of hats

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

olon

-0.70

 Recover

-0.67

venant

-0.65

rans

-0.61

 Document

-0.60

eway

-0.59

 Citation

-0.59

 Residential

-0.58

ara

-0.58

LR

-0.58

POSITIVE LOGITS

 hats

4.11

hat

2.16

 Hats

2.11

 helmets

1.72

 shirts

1.61

 jackets

1.59

 coats

1.59

 masks

1.57

 costumes

1.55

Hat

1.54

Activations Density 0.020%

mentions of hats

mentions of hats

No Comments

No Known Activations