INDEX

Explanations

numbers in textual formats

oai_token-act-pair · gpt-3.5-turbo

occurrences of various numerical references and related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

hips

-0.89

loo

-0.83

WARD

-0.70

wards

-0.69

ioned

-0.68

 Denis

-0.67

fully

-0.66

 Hilton

-0.63

lain

-0.63

 Leopard

-0.61

POSITIVE LOGITS

eral

1.05

ero

1.04

posium

1.03

pty

1.03

mus

0.95

phony

0.92

bs

0.92

pt

0.87

asm

0.86

urg

0.86

Activations Density 0.042%

numbers in textual formats

occurrences of various numerical references and related terms

No Comments

No Known Activations