INDEX

Explanations

email-related terms and acronyms

oai_token-act-pair · gpt-3.5-turbo

elements related to email usage and regulatory terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 tremend

-0.77

 rall

-0.68

 pione

-0.65

 ende

-0.64

 notor

-0.63

 eleph

-0.63

 occas

-0.62

 pecul

-0.62

 reluct

-0.62

 millenn

-0.62

POSITIVE LOGITS

igrant

0.76

ogenesis

0.76

akedown

0.72

emption

0.71

mediated

0.71

asis

0.71

igrants

0.69

aneers

0.68

liction

0.67

ergy

0.66

Activations Density 0.230%

email-related terms and acronyms

elements related to email usage and regulatory terms

No Comments

No Known Activations