INDEX

Explanations

phrases related to unique or important items or concepts

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

Downloadha

-0.88

raints

-0.79

UFF

-0.74

æ©Ł

-0.72

arium

-0.71

orsi

-0.70

ITNESS

-0.66

ampunk

-0.66

maxwell

-0.66

apons

-0.66

POSITIVE LOGITS

handedly

1.49

ton

1.20

player

1.03

 digit

0.98

 digits

0.98

minded

0.96

 minded

0.96

 sided

0.96

digit

0.87

 handed

0.84

Activations Density 0.400%

phrases related to unique or important items or concepts

No Comments

No Known Activations