INDEX

Explanations

journalistic terms or activities related to reporting and editing

oai_token-act-pair · gpt-3.5-turbo

references to journalistic activities and reporting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ewitness

-0.74

omet

-0.72

imus

-0.71

ESCO

-0.68

coni

-0.66

oping

-0.64

utsche

-0.64

ification

-0.63

hift

-0.63

hens

-0.63

POSITIVE LOGITS

 excellence

0.97

 prowess

0.87

 techniques

0.82

 Excellence

0.79

 Standards

0.78

 standards

0.77

 styles

0.75

 style

0.74

 chops

0.73

 Practices

0.72

Activations Density 0.138%

journalistic terms or activities related to reporting and editing

references to journalistic activities and reporting

No Comments

No Known Activations