INDEX

Explanations

words related to lawyers

oai_token-act-pair · gpt-3.5-turbo

instances of the word "lawyer."

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

 isot

-0.67

 disk

-0.66

 tanks

-0.63

 memory

-0.61

 Persian

-0.60

 curs

-0.60

 offensive

-0.60

 disks

-0.59

wra

-0.59

 stamps

-0.59

POSITIVE LOGITS

yer

4.65

yers

2.48

ye

1.18

iard

1.12

suit

1.08

YE

1.07

yard

1.05

yre

1.04

yo

1.01

maker

0.99

Activations Density 0.017%

words related to lawyers

instances of the word "lawyer."

No Comments

No Known Activations