INDEX

Explanations

instances of the word "to" followed by another word or phrase

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

phrases indicating intentions or future actions

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

Downloadha

-0.76

heit

-0.73

listed

-0.71

 è£ıè¦ļéĨĴ

-0.69

cells

-0.69

 Leban

-0.69

abiding

-0.68

hea

-0.68

felt

-0.67

POSITIVE LOGITS

 unveil

1.11

 intensify

0.99

 revisit

0.98

 replicate

0.97

 capitalize

0.97

 introduce

0.97

sue

0.96

 emulate

0.96

 spend

0.95

 continue

0.94

Activations Density 0.186%

instances of the word "to" followed by another word or phrase

phrases indicating intentions or future actions

No Comments

No Known Activations