INDEX

Explanations

names and numbers

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

aido

-0.68

natureconservancy

-0.68

Shares

-0.66

 Reborn

-0.63

uca

-0.63

KEY

-0.62

pak

-0.61

zynski

-0.60

ONEY

-0.59

grave

-0.58

POSITIVE LOGITS

tenance

1.12

llor

0.90

acters

0.89

osterone

0.85

lihood

0.83

quartered

0.81

riors

0.79

clusive

0.76

iste

0.75

ruction

0.75

Activations Density 3.434%

names and numbers

No Comments

No Known Activations