INDEX

Explanations

phrases indicating a contrast or contradiction

oai_token-act-pair · gpt-3.5-turbo

the word "how" 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 @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

uthor

-0.65

agonists

-0.63

 Grail

-0.62

lehem

-0.61

 Guer

-0.58

iculture

-0.57

ception

-0.57

agonist

-0.56

isher

-0.56

 Yard

-0.56

POSITIVE LOGITS

soever

1.08

HCR

0.86

beit

0.86

ever

0.82

ls

0.82

ling

0.81

ells

0.77

itzer

0.76

 much

0.75

 MUCH

0.74

Activations Density 0.084%

phrases indicating a contrast or contradiction

the word "how" in various contexts

No Comments

No Known Activations