INDEX

Explanations

words related to evaluating something against a particular standard

oai_token-act-pair · gpt-3.5-turbo

references to standards and regulations

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

 renewed

-0.65

 rejo

-0.64

 activity

-0.64

 afar

-0.63

 revenge

-0.62

 defeated

-0.61

ads

-0.59

 nervous

-0.59

 vengeance

-0.59

 overcoming

-0.58

POSITIVE LOGITS

standard

4.06

Standard

1.99

 standard

1.85

 Standard

1.52

normal

1.39

basic

1.28

default

1.28

traditional

1.26

average

1.23

minimum

1.22

Activations Density 0.013%

words related to evaluating something against a particular standard

references to standards and regulations

No Comments

No Known Activations