INDEX

Explanations

specific letters or abbreviations, notably "B."

oai_token-act-pair · gpt-3.5-turbo

references to sports teams or organizations, particularly those with the initial 'B'

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

Not in Any Lists

No Comments

Negative Logits

 Duterte

-0.70

 Sahara

-0.70

anmar

-0.69

mum

-0.68

 Divinity

-0.67

 Tablet

-0.66

Mum

-0.66

 Cheong

-0.66

 Idle

-0.66

 Tanz

-0.65

POSITIVE LOGITS

FC

0.99

BA

0.95

HL

0.90

NBA

0.90

LCS

0.90

BL

0.90

League

0.89

FL

0.89

BU

0.87

NFL

0.82

Activations Density 0.086%

specific letters or abbreviations, notably "B."

references to sports teams or organizations, particularly those with the initial 'B'

No Comments

No Known Activations