INDEX

Explanations

references to quantities or numbers

oai_token-act-pair · gpt-3.5-turbo

phrases related to numerical data or lists

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 misunder

-0.66

 agre

-0.64

 fert

-0.64

agher

-0.61

 arrang

-0.60

uala

-0.59

 advert

-0.59

hunt

-0.58

vill

-0.57

 depot

-0.57

POSITIVE LOGITS

icial

0.87

ources

0.77

 course

0.68

neys

0.68

course

0.66

 consecutive

0.65

uits

0.63

 sorts

0.60

eworks

0.60

 Southwest

0.60

Activations Density 0.131%

references to quantities or numbers

phrases related to numerical data or lists

No Comments

No Known Activations