INDEX

Explanations

phrases related to current events and news stories

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

 seiz

-0.67

pora

-0.66

 subsequ

-0.60

 challeng

-0.60

antage

-0.59

luster

-0.57

der

-0.57

 necks

-0.57

antis

-0.57

 Pett

-0.56

POSITIVE LOGITS

ifiable

1.39

ifications

1.31

if

0.99

ified

0.96

ification

0.96

itia

0.95

ifi

0.94

 kidding

0.93

IFIED

0.91

ices

0.90

Activations Density 0.321%

phrases related to current events and news stories

No Comments

No Known Activations