INDEX

Explanations

phrases related to technical instructions or commands

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

instances of the word "try" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

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_post

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

 conduc

-0.59

 metic

-0.58

LET

-0.58

 Beir

-0.57

 promot

-0.56

aditional

-0.56

 advoc

-0.56

 mathemat

-0.56

 Kardash

-0.56

 enthusi

-0.56

POSITIVE LOGITS

 again

1.19

 Again

0.83

Again

0.74

again

0.74

 twice

0.65

ceed

0.64

 harder

0.64

 another

0.64

ause

0.63

ogg

0.62

Activations Density 0.010%

phrases related to technical instructions or commands

instances of the word "try" in various contexts

No Comments

No Known Activations