INDEX

Explanations

words related to names or titles of individuals

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

references to historical figures and events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

taboola

-0.73

WD

-0.68

Story

-0.67

âľ

-0.66

achy

-0.64

Ops

-0.64

quad

-0.64

[/

-0.63

pour

-0.63

vc

-0.62

POSITIVE LOGITS

 Adolf

0.95

 famed

0.89

 royalty

0.88

 Julius

0.85

 Founding

0.84

 Abraham

0.83

 Winston

0.83

 landmarks

0.82

 antiquity

0.82

 Queen

0.82

Activations Density 0.342%

words related to names or titles of individuals

references to historical figures and events

No Comments

No Known Activations