INDEX

Explanations

references to hand-related activities or objects

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

vell

-0.76

 Tonight

-0.74

ashtra

-0.74

 RAND

-0.74

 Atmosp

-0.72

actionDate

-0.72

idian

-0.70

 Neigh

-0.69

ctory

-0.69

nces

-0.68

POSITIVE LOGITS

bag

1.21

maid

1.18

picked

1.09

shake

1.06

crafted

1.06

bags

1.03

book

0.99

ily

0.95

ker

0.93

books

0.92

Activations Density 0.055%

references to hand-related activities or objects

No Comments

No Known Activations