INDEX

Explanations

abbreviations or acronyms with periods in between

oai_token-act-pair · gpt-3.5-turbo

punctuation marks

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 advis

-0.82

 snapshot

-0.72

 induct

-0.70

 charact

-0.69

 retrospective

-0.69

 transact

-0.68

 enthusi

-0.68

 guiActiveUn

-0.68

ihar

-0.66

 taco

-0.66

POSITIVE LOGITS

 Afterwards

0.79

assetsadobe

0.76

 Browne

0.73

 Retrieved

0.73

Org

0.70

icio

0.70

jpg

0.70

 Geological

0.69

 Freeman

0.69

Og

0.68

Activations Density 0.470%

abbreviations or acronyms with periods in between

punctuation marks

No Comments

No Known Activations