INDEX

Explanations

phrases related to exerting force or pressure on someone or something

oai_token-act-pair · gpt-3.5-turbo

references to compulsion or coercion

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

olate

-0.67

 Threat

-0.66

 Blueprint

-0.63

 Echoes

-0.63

terday

-0.62

SEA

-0.62

 Billboard

-0.62

Pru

-0.60

 Laksh

-0.60

 Emerging

-0.59

POSITIVE LOGITS

 wedge

0.71

ocker

0.70

DERR

0.70

 conformity

0.68

 removal

0.66

emon

0.65

 withdrawals

0.64

 cancellation

0.63

 havoc

0.62

onto

0.61

Activations Density 0.137%

phrases related to exerting force or pressure on someone or something

references to compulsion or coercion

No Comments

No Known Activations