INDEX

Explanations

terms that indicate a sense of danger or risk

oai_token-act-pair · gpt-3.5-turbo

instances of the word "threatening."

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

icion

-0.72

amel

-0.72

 Selection

-0.71

DX

-0.65

Tab

-0.65

omics

-0.64

 Couch

-0.62

 refresh

-0.62

HH

-0.62

 Feet

-0.62

POSITIVE LOGITS

 threatening

3.49

threatening

2.30

 threaten

2.18

 menacing

2.06

 threatened

2.06

 threats

2.04

threat

1.85

 harassing

1.84

 threatens

1.79

 threat

1.79

Activations Density 0.013%

terms that indicate a sense of danger or risk

instances of the word "threatening."

No Comments

No Known Activations