INDEX

Explanations

phrases related to thoughts, ideas, and opinions

oai_token-act-pair · gpt-3.5-turbo

expressions of thoughts or opinions about various subjects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

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

è¦ļéĨĴ

-0.75

ãĤ¬

-0.72

ãĤ´ãĥ³

-0.70

ãĤº

-0.68

 contention

-0.67

WAR

-0.66

 Nationwide

-0.66

adra

-0.64

OURCE

-0.63

 Regions

-0.62

POSITIVE LOGITS

'd

1.00

 might

0.99

 sounded

0.92

 prudent

0.87

 ought

0.84

 joking

0.83

odd

0.82

might

0.82

 funny

0.79

 hilarious

0.77

Activations Density 0.112%

phrases related to thoughts, ideas, and opinions

expressions of thoughts or opinions about various subjects

No Comments

No Known Activations