INDEX

Explanations

dialogue and actions related to characters' emotional experiences

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 Carson

-0.96

espie

-0.93

 bolted

-0.88

 helicop

-0.85

 Webb

-0.84

 loaf

-0.83

 ranc

-0.81

Colorado

-0.81

 Lyons

-0.80

 Nebraska

-0.78

POSITIVE LOGITS

 manga

1.52

 anime

1.45

 ãĢĮ

1.44

 Anime

1.41

 Yamato

1.38

ãĢİ

1.38

ãĢĲ

1.35

æ

1.35

âĺĨ

1.34

 ãĥ

1.33

Activations Density 0.474%

dialogue and actions related to characters' emotional experiences

No Comments

No Known Activations