INDEX

Explanations

specific locations and events mentioned in a text

oai_token-act-pair · gpt-3.5-turbo

references to specific locations and events related to crime or violence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

cipled

-0.73

laus

-0.65

cellaneous

-0.59

detail

-0.55

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

-0.54

enture

-0.54

gov

-0.53

ertodd

-0.53

areth

-0.53

yne

-0.51

POSITIVE LOGITS

 badge

0.76

 moniker

0.67

 persona

0.64

onto

0.63

 slate

0.63

 salute

0.61

 barrier

0.61

 flag

0.59

 masterpiece

0.59

cra

0.59

Activations Density 2.774%

specific locations and events mentioned in a text

references to specific locations and events related to crime or violence

No Comments

No Known Activations