INDEX

Explanations

information related to locations or places, specifically with the word "Hang" or "Hangar"

oai_token-act-pair · gpt-3.5-turbo

references to the word "Hang" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 violet

-0.79

 serpent

-0.75

orc

-0.72

ierre

-0.68

Murray

-0.65

 palette

-0.63

eur

-0.63

Ã«

-0.63

£ı

-0.62

na

-0.62

POSITIVE LOGITS

 Hang

3.85

hang

2.30

 hangar

1.21

 Kang

1.20

 Shanghai

1.19

 Shen

1.13

 Shang

1.12

 Hung

1.08

 Mandarin

1.06

 Hamburg

1.05

Activations Density 0.028%

information related to locations or places, specifically with the word "Hang" or "Hangar"

references to the word "Hang" and related terms

No Comments

No Known Activations