INDEX

Explanations

questions or requests for information

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

 Confeder

-0.67

 Fried

-0.65

Est

-0.64

 Mong

-0.64

 mars

-0.64

herence

-0.63

ources

-0.62

 Siber

-0.58

 Ange

-0.58

oos

-0.58

POSITIVE LOGITS

 repeatedly

0.83

 questions

0.77

by

0.76

 aback

0.72

naire

0.71

BY

0.70

recated

0.69

 question

0.68

 choose

0.67

naires

0.66

Activations Density 0.053%

questions or requests for information

No Comments

No Known Activations