INDEX

Explanations

instances where someone or something is suspected or accused of something negative

oai_token-act-pair · gpt-3.5-turbo

phrases relating to the concept of being responsible for actions or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

istic

-0.75

ism

-0.68

cki

-0.68

Pwr

-0.67

ander

-0.66

ERC

-0.63

District

-0.62

aires

-0.61

era

-0.61

aire

-0.60

POSITIVE LOGITS

track

0.74

âĸ¬âĸ¬

0.73

loading

0.72

coat

0.72

 bars

0.71

¬

0.69

urat

0.68

aan

0.65

lined

0.65

shore

0.65

Activations Density 0.016%

instances where someone or something is suspected or accused of something negative

phrases relating to the concept of being responsible for actions or events

No Comments

No Known Activations