INDEX

Explanations

references to harm and physics-related terms

oai_token-act-pair · gpt-3.5-turbo

terms related to harm and its effects

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

Ö¼

-0.87

onde

-0.74

 stakes

-0.73

 toes

-0.70

 ducks

-0.67

 Monteneg

-0.66

 thumbs

-0.65

 eyed

-0.65

 craw

-0.64

 gravel

-0.64

POSITIVE LOGITS

harm

3.17

phys

1.39

 physi

1.34

 Phys

1.32

 Harm

1.28

 pharmac

1.23

Phys

1.20

alter

1.13

aber

1.07

ulla

1.03

Activations Density 0.056%

references to harm and physics-related terms

terms related to harm and its effects

No Comments

No Known Activations