INDEX

Explanations

phrases related to subscribing, supporting, and contributing to a news organization

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

âĢİ

-0.68

ndum

-0.65

ovember

-0.65

 rehearsal

-0.63

imer

-0.61

[+

-0.59

 Prob

-0.57

uthor

-0.57

abal

-0.57

geist

-0.57

POSITIVE LOGITS

entimes

0.68

APD

0.67

interstitial

0.65

 è£ıè

0.63

independence

0.62

 domain

0.62

é¾įåĸļå£«

0.62

osion

0.61

æ©

0.61

 domains

0.60

Activations Density 0.564%

phrases related to subscribing, supporting, and contributing to a news organization

No Comments

No Known Activations