INDEX

Explanations

quantitative measurements or values being compared to a specific threshold

oai_token-act-pair · gpt-3.5-turbo

phrases indicating quality or condition, particularly positive assessments

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rely

-0.69

omore

-0.67

apixel

-0.67

alien

-0.64

acity

-0.64

iferation

-0.59

opol

-0.57

eway

-0.57

algia

-0.57

ovo

-0.56

POSITIVE LOGITS

 overdue

1.04

spring

0.99

 underway

0.99

 vers

0.95

 intention

0.93

 beyond

0.91

 acquainted

0.90

 behaved

0.90

 suited

0.89

 aware

0.88

Activations Density 0.038%

quantitative measurements or values being compared to a specific threshold

phrases indicating quality or condition, particularly positive assessments

No Comments

No Known Activations