INDEX

Explanations

references to photography or filming techniques involving taking multiple shots or exposures and combining them

oai_token-act-pair · gpt-3.5-turbo

references to "shots" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ŃĶ

-0.86

ordinate

-0.82

lict

-0.76

ebook

-0.74

doms

-0.73

rab

-0.72

gres

-0.69

Case

-0.68

development

-0.68

rador

-0.67

POSITIVE LOGITS

 shots

1.36

 shot

1.09

 Shots

1.07

shots

0.92

hooting

0.90

 shoots

0.90

 Fired

0.88

 shoot

0.86

shot

0.86

 fired

0.83

Activations Density 0.012%

references to photography or filming techniques involving taking multiple shots or exposures and combining them

references to "shots" in various contexts

No Comments

No Known Activations