INDEX

Explanations

phrases related to investigations or matters being looked into

oai_token-act-pair · gpt-3.5-turbo

references to ongoing investigations or issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥ³ãĤ¸

-0.97

ARP

-0.87

©¶æ

-0.82

NetMessage

-0.81

ãĥ¼ãĥĨ

-0.76

eele

-0.72

KI

-0.71

cki

-0.71

asio

-0.70

iens

-0.69

POSITIVE LOGITS

ional

0.94

 matters

0.88

manship

0.86

 relating

0.84

 pertaining

0.84

 involving

0.84

forward

0.82

 surrounding

0.80

enance

0.80

 concerning

0.80

Activations Density 0.017%

phrases related to investigations or matters being looked into

references to ongoing investigations or issues

No Comments

No Known Activations