INDEX

Explanations

instances where the word "mistake" is present, regardless of context

oai_token-act-pair · gpt-3.5-turbo

terms related to errors or mistakes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Hughes

-0.81

 Kaufman

-0.74

 Levine

-0.73

 homosexuality

-0.72

 veget

-0.71

 Arnold

-0.69

 animate

-0.68

 Subway

-0.66

 injected

-0.66

 eggs

-0.66

POSITIVE LOGITS

mist

3.52

Mist

2.52

 Mist

2.25

search

1.23

 Mistress

1.15

misc

1.11

 Ranger

1.09

 slightest

1.04

icion

1.01

ISS

1.01

Activations Density 0.055%

instances where the word "mistake" is present, regardless of context

terms related to errors or mistakes

No Comments

No Known Activations