INDEX

Explanations

information related to different individuals and their backgrounds

oai_token-act-pair · gpt-3.5-turbo

references to specific identities and roles of individuals

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

hower

-0.63

 disperse

-0.59

 incent

-0.58

 reinforcements

-0.58

 VIDEOS

-0.55

 GEAR

-0.55

effic

-0.53

 Transcript

-0.52

 Repeat

-0.52

 scramble

-0.51

POSITIVE LOGITS

anka

0.59

 antit

0.54

itar

0.53

iator

0.53

à¹

0.53

 Lana

0.52

 Lazarus

0.51

Must

0.51

ank

0.51

onite

0.51

Activations Density 1.712%

information related to different individuals and their backgrounds

references to specific identities and roles of individuals

No Comments

No Known Activations