INDEX

Explanations

encoded structured data or specific markers within the text

oai_token-act-pair · gpt-3.5-turbo

indicators of upcoming events or ongoing activities

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 eleph

-0.86

 occas

-0.83

 hemor

-0.82

 citiz

-0.80

 tremend

-0.77

cius

-0.74

 proport

-0.73

 unnecess

-0.70

aditional

-0.68

 exting

-0.66

POSITIVE LOGITS

Liverpool

0.90

Chelsea

0.90

England

0.87

Sky

0.86

Brazil

0.80

Manager

0.78

Republic

0.77

Manchester

0.77

Rober

0.76

Bay

0.76

Activations Density 0.173%

encoded structured data or specific markers within the text

indicators of upcoming events or ongoing activities

No Known Activations