INDEX

Explanations

proper names, particularly those related to "Gan" and "Kok"

oai_token-act-pair · gpt-3.5-turbo

mentions of specific names, particularly focusing on the name "Gan" and other related individuals

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 seeker

-0.77

anwhile

-0.74

sburgh

-0.70

 COURT

-0.66

 BART

-0.64

 Cassandra

-0.63

 Patriot

-0.62

 Parish

-0.61

 magnitude

-0.60

 referrals

-0.60

POSITIVE LOGITS

omal

1.05

bare

1.04

esh

1.04

ryu

0.99

omen

0.97

zhou

0.97

Gan

0.97

zu

0.92

zeb

0.92

uin

0.91

Activations Density 0.011%

proper names, particularly those related to "Gan" and "Kok"

mentions of specific names, particularly focusing on the name "Gan" and other related individuals

No Comments

No Known Activations