INDEX

Explanations

dates and related terms like years

oai_token-act-pair · gpt-3.5-turbo

formatted date and time references

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

 Moonlight

-0.68

erville

-0.64

avorite

-0.62

 Fantasy

-0.61

 clipboard

-0.61

 Strongh

-0.59

 Facilities

-0.57

riber

-0.56

 mate

-0.56

inence

-0.56

POSITIVE LOGITS

 partName

0.90

DragonMagazine

0.81

)</

0.73

msg

0.72

uments

0.68

ument

0.66

sv

0.66

owa

0.66

govtrack

0.66

uploads

0.65

Activations Density 0.047%

dates and related terms like years

formatted date and time references

No Comments

No Known Activations