INDEX

Explanations

references to sizes or scales in the context of various types of measurements

oai_token-act-pair · gpt-3.5-turbo

conjunctions and prepositions indicating relationships or connections

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

eries

-0.82

ardless

-0.76

 Voices

-0.73

 Pigs

-0.70

 VIDEOS

-0.70

Images

-0.69

 Myster

-0.69

Apps

-0.68

fits

-0.68

 Hearts

-0.67

POSITIVE LOGITS

 intermediate

1.16

 upper

0.99

 late

0.99

 high

0.98

 middle

0.95

 senior

0.94

 advanced

0.92

mid

0.89

 junior

0.86

 medium

0.86

Activations Density 0.101%

references to sizes or scales in the context of various types of measurements

conjunctions and prepositions indicating relationships or connections

No Comments

No Known Activations