INDEX

Explanations

the name "Gwen"

oai_token-act-pair · gpt-3.5-turbo

instances of the name "Wen."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ĺħ

-0.75

Ħ¢

-0.72

 Seym

-0.70

aminer

-0.66

 practicable

-0.65

 susceptibility

-0.65

osate

-0.65

 plural

-0.64

 cultured

-0.63

ô

-0.62

POSITIVE LOGITS

wen

1.35

burn

1.03

mere

0.97

borough

0.94

 Kenobi

0.92

igan

0.91

pool

0.86

igans

0.86

boro

0.84

jen

0.84

Activations Density 0.003%

the name "Gwen"

instances of the name "Wen."

No Comments

No Known Activations