INDEX

Explanations

references to assistant roles and titles in legal and government contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

torch.float32

Hook Point

blocks.8.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

tumblr

-0.78

ebook

-0.74

 foss

-0.73

naires

-0.72

 buckle

-0.72

 pasture

-0.71

 shave

-0.71

ighting

-0.70

ograph

-0.69

umblr

-0.69

POSITIVE LOGITS

 Director

1.34

 Assistant

1.25

 Coordinator

1.23

 Editor

1.21

 Secretary

1.19

 Attorney

1.16

 Minister

1.15

 Principal

1.15

 Analyst

1.14

 Investigator

1.13

Activations Density 0.025%

references to assistant roles and titles in legal and government contexts

No Comments

No Known Activations