INDEX

Explanations

content related to editing or editing instructions in a document

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

Ń·

-0.82

acies

-0.73

İĭ

-0.67

aido

-0.67

¿½

-0.66

ãĥ¼ãĥĨãĤ£

-0.65

 undermin

-0.64

aughters

-0.64

ĵĺ

-0.64

etsy

-0.63

POSITIVE LOGITS

TPS

0.80

 onwards

0.72

arthed

0.66

sbm

0.65

semble

0.64

 Cheong

0.64

 Panc

0.63

},{"

0.62

addons

0.62

+)

0.62

Activations Density 0.743%

content related to editing or editing instructions in a document

No Comments

No Known Activations