INDEX

Explanations

references to a specific term or name "opa" with different activation levels

oai_token-act-pair · gpt-3.5-turbo

references to "Hoopa" and "ARP" related terms or abbreviations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

des

-0.73

een

-0.72

washer

-0.69

 Wick

-0.69

Rust

-0.66

 Wolver

-0.65

must

-0.65

rum

-0.65

coat

-0.64

need

-0.64

POSITIVE LOGITS

OPA

1.30

vernment

1.08

agall

1.03

odcast

0.99

phrine

0.94

merce

0.91

opa

0.91

etsk

0.89

yright

0.88

mercial

0.88

Activations Density 0.006%

references to a specific term or name "opa" with different activation levels

references to "Hoopa" and "ARP" related terms or abbreviations

No Comments

No Known Activations