INDEX

Explanations

proper nouns, likely related to names of people or places

oai_token-act-pair · gpt-3.5-turbo

references to surveillance and public exposure

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

sylv

-0.52

farious

-0.51

rium

-0.49

awa

-0.48

Loading

-0.48

chuk

-0.47

 Vaugh

-0.45

ploma

-0.44

romeda

-0.44

igmat

-0.42

POSITIVE LOGITS

 largeDownload

0.51

bably

0.49

pmwiki

0.47

 è£ıè

0.47

 Preferred

0.47

 âĢº

0.46

fecture

0.45

ARM

0.45

disk

0.44

 brute

0.43

Activations Density 1.779%

proper nouns, likely related to names of people or places

references to surveillance and public exposure

No Comments

No Known Activations