INDEX

Explanations

mentions related to television, television shows, and television personalities

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

urion

-0.83

etsy

-0.74

cy

-0.69

lein

-0.67

uria

-0.66

vironment

-0.64

ixir

-0.64

orget

-0.63

ulkan

-0.62

paper

-0.62

POSITIVE LOGITS

 broadcasts

1.20

 aired

1.19

 airing

1.17

 television

1.14

TV

1.11

 viewers

1.06

 viewership

1.06

 commercials

1.05

 Channel

1.05

 premie

1.05

Activations Density 4.520%

mentions related to television, television shows, and television personalities

No Comments

No Known Activations