INDEX

Explanations

proper nouns related to names or entities

oai_token-act-pair · gpt-3.5-turbo

references to specific characters or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 plateau

-0.78

berto

-0.69

oust

-0.69

 Rousse

-0.68

Flor

-0.68

 bottleneck

-0.68

 bass

-0.68

 limestone

-0.67

 decomp

-0.66

 fever

-0.64

POSITIVE LOGITS

 Ally

3.92

Om

1.47

 Nationwide

1.16

 Allies

1.08

 Enemy

1.05

Hed

1.01

 Mutual

0.97

 Bless

0.89

 Robot

0.88

 Shades

0.86

Activations Density 0.034%

proper nouns related to names or entities

references to specific characters or entities

No Comments

No Known Activations