INDEX

Explanations

mentions of entities or products identified by an alphanumeric naming convention, possibly related to technology or media

oai_token-act-pair · gpt-3.5-turbo

references to television or media series

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

eff

-0.67

 Mane

-0.59

gee

-0.58

 Newport

-0.58

 selfies

-0.58

Kem

-0.57

 Brune

-0.55

 democr

-0.55

 emanc

-0.55

gib

-0.55

POSITIVE LOGITS

Series

3.91

series

3.72

 Series

1.97

 series

1.52

Season

1.36

Plot

1.23

Episode

1.21

episode

1.17

Film

0.96

 trilogy

0.94

Activations Density 0.013%

mentions of entities or products identified by an alphanumeric naming convention, possibly related to technology or media

references to television or media series

No Comments

No Known Activations