INDEX

Explanations

references to substances or drugs, especially in the context of abuse or medical treatment

oai_token-act-pair · gpt-3.5-turbo

references to substance abuse and related issues

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

ITAL

-0.71

 Zimmer

-0.70

ãĥ¼ãĤ¯

-0.69

DIR

-0.69

OY

-0.66

aldi

-0.66

 Ellison

-0.65

ros

-0.64

 Blazers

-0.64

 Fior

-0.62

POSITIVE LOGITS

fulness

0.99

 substance

0.96

 substances

0.92

abuse

0.89

amphetamine

0.86

cohol

0.85

 Substance

0.84

 abuser

0.84

 abusers

0.83

iveness

0.83

Activations Density 0.012%

references to substances or drugs, especially in the context of abuse or medical treatment

references to substance abuse and related issues

No Comments

No Known Activations