INDEX

Explanations

references to the specific word "Duke."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Du" followed by a numerical designation or reference

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

clinton

-0.65

enance

-0.64

nikov

-0.62

giving

-0.61

ãģĨ

-0.61

èĢħ

-0.60

 WRITE

-0.60

 unaffected

-0.60

 Debor

-0.59

âĶģ

-0.58

POSITIVE LOGITS

Pont

1.29

pless

0.97

athlon

0.96

opoly

0.96

isine

0.90

arte

0.89

plin

0.87

xton

0.85

ress

0.85

pe

0.83

Activations Density 0.020%

references to the specific word "Duke."

instances of the word "Du" followed by a numerical designation or reference

No Comments

No Known Activations