INDEX

Explanations

phrases related to the concept of writing or appearing as someone else, known as ghostwriting

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

erity

-0.82

Effective

-0.78

ahime

-0.77

onian

-0.70

unes

-0.70

baugh

-0.69

nice

-0.68

enegger

-0.67

tics

-0.67

effective

-0.66

POSITIVE LOGITS

busters

1.08

 haunting

0.99

buster

0.91

written

0.89

writer

0.87

writing

0.82

luster

0.79

 hunters

0.78

 haunt

0.78

kered

0.77

Activations Density 0.088%

phrases related to the concept of writing or appearing as someone else, known as ghostwriting

No Comments

No Known Activations