INDEX

Explanations

abbreviated names of organizations or institutions followed by actions

oai_token-act-pair · gpt-3.5-turbo

references to governmental and organizational entities

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

ãĥĥãĥī

-0.75

ãĥ´ãĤ¡

-0.74

 answ

-0.73

abilia

-0.72

wcs

-0.72

ãĥĦ

-0.72

ãĤ©

-0.71

icter

-0.70

ãĥ¥

-0.69

zers

-0.68

POSITIVE LOGITS

 Committees

0.67

 Protective

0.64

 Syndicate

0.62

 tasked

0.60

 Guild

0.60

present

0.60

 Cooperative

0.59

 Mines

0.58

semble

0.58

 Construct

0.58

Activations Density 0.106%

abbreviated names of organizations or institutions followed by actions

references to governmental and organizational entities

No Comments

No Known Activations