INDEX

Explanations

words that contain the specific sequence "rom" with a high activation value

oai_token-act-pair · gpt-3.5-turbo

instances of the word "rom" in various contexts

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

Notes

-0.72

Ĵ

-0.70

ŃĶ

-0.67

 Citation

-0.67

 Reserve

-0.67

 Jackets

-0.64

 attribution

-0.63

fav

-0.62

 need

-0.61

 Venezuela

-0.61

POSITIVE LOGITS

rom

1.24

antic

1.12

antically

1.04

astery

1.01

thal

1.00

ancers

1.00

onse

0.95

agnetic

0.95

antics

0.93

orph

0.92

Activations Density 0.008%

words that contain the specific sequence "rom" with a high activation value

instances of the word "rom" in various contexts

No Comments

No Known Activations