INDEX

Explanations

proper names and technical terms related to different contexts, such as individuals, locations, and technology

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals and names

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

lain

-0.80

arily

-0.75

wagen

-0.74

ularity

-0.71

ULAR

-0.71

iated

-0.71

ierra

-0.69

liga

-0.69

ibal

-0.69

 confir

-0.69

POSITIVE LOGITS

 Rodney

1.17

 McKay

0.87

Smy

0.84

Frames

0.81

tons

0.81

neys

0.79

 Quincy

0.78

 Crow

0.76

 Danger

0.76

 Avery

0.75

Activations Density 0.023%

proper names and technical terms related to different contexts, such as individuals, locations, and technology

references to specific individuals and names

No Comments

No Known Activations