INDEX

Explanations

phrases related to legal judgments or criminal investigations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "found" and associated legal outcomes or verdicts

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

dos

-0.74

 Confederation

-0.67

 Inqu

-0.67

idth

-0.66

ription

-0.65

gemony

-0.65

 obser

-0.62

eatures

-0.62

 Drill

-0.61

iann

-0.61

POSITIVE LOGITS

 guilty

1.56

 liable

1.19

 Guilty

1.15

 unfit

1.09

 ineligible

1.07

 dead

1.06

 hanged

1.05

 culp

1.02

 responsible

1.00

 incompetent

1.00

Activations Density 0.048%

phrases related to legal judgments or criminal investigations

instances of the word "found" and associated legal outcomes or verdicts

No Comments

No Known Activations