INDEX

Explanations

explicit references to specific events or incidents, potentially with locations and details

oai_token-act-pair · gpt-3.5-turbo

articles and determiners

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

 legions

-0.72

osponsors

-0.72

âĹ¼

-0.70

attacks

-0.67

 essays

-0.67

 feats

-0.67

 armies

-0.66

ships

-0.66

 alliances

-0.65

 scandals

-0.64

POSITIVE LOGITS

 neighbor

0.95

 woman

0.95

 bystand

0.92

 handgun

0.91

 passer

0.89

 neighbour

0.86

 disturbance

0.85

 burglary

0.83

 dispatcher

0.83

 homeowner

0.82

Activations Density 0.184%

explicit references to specific events or incidents, potentially with locations and details

articles and determiners

No Comments

No Known Activations