INDEX

Explanations

references to physical injuries and medical conditions

oai_token-act-pair · gpt-3.5-turbo

references to injuries and wounds

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

essions

-0.75

oggle

-0.73

Ü

-0.71

urch

-0.69

rador

-0.69

oS

-0.68

ramid

-0.67

ordan

-0.67

iazep

-0.66

ondon

-0.65

POSITIVE LOGITS

 wounds

1.34

 inflicted

1.28

 scars

0.99

 wound

0.86

 wounded

0.83

 wounding

0.82

 suffered

0.81

 healed

0.81

 pierced

0.79

 incurred

0.78

Activations Density 0.020%

references to physical injuries and medical conditions

references to injuries and wounds

No Comments

No Known Activations