INDEX

Explanations

names, terms, and phrases associated with chess and various languages

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

eatures

-0.77

 viability

-0.71

earchers

-0.71

 sponsors

-0.71

ulators

-0.70

 concede

-0.68

 sponsorship

-0.68

 stake

-0.68

 athlet

-0.66

 implants

-0.66

POSITIVE LOGITS

ÙĨ

1.64

Ø§

1.56

Ùĩ

1.50

²

1.49

Ù

1.48

¹

1.46

´

1.44

ļ

1.42

ÙĬ

1.42

Ø§Ø

1.41

Activations Density 1.764%

names, terms, and phrases associated with chess and various languages

No Comments

No Known Activations