INDEX

Explanations

mentions of specific locations and actions related to incidents or events

oai_token-act-pair · gpt-3.5-turbo

references to individuals and events related to crime or distress

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ACTIONS

-0.71

nesday

-0.71

rates

-0.68

ichever

-0.67

 ensures

-0.67

razil

-0.67

 Cosponsors

-0.66

utsche

-0.66

é¾įå¥ĳå£«

-0.63

ivalry

-0.63

POSITIVE LOGITS

 lying

1.31

 slumped

1.14

 asleep

1.07

 hiding

1.06

 hanging

1.06

 sitting

1.04

 dangling

1.04

 clinging

1.02

 missing

1.02

 staring

1.02

Activations Density 0.344%

mentions of specific locations and actions related to incidents or events

references to individuals and events related to crime or distress

No Comments

No Known Activations