INDEX

Explanations

references to individuals who are former or ex-related to specific roles or professions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

uden

-0.90

Zup

-0.80

 shading

-0.78

 dances

-0.77

oglu

-0.71

 weighted

-0.67

 largeDownload

-0.66

 underlying

-0.65

kson

-0.64

 emoji

-0.63

POSITIVE LOGITS

girlfriend

1.34

husband

1.31

wife

1.17

members

1.16

employ

1.15

member

1.14

Soviet

1.13

military

1.10

commun

1.09

marine

1.07

Activations Density 0.010%

references to individuals who are former or ex-related to specific roles or professions

No Comments

No Known Activations

references to individuals who are former or ex-related to specific roles or professions

No Comments

No Known Activations