INDEX

Explanations

phrases related to news articles and journalism

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

ļé

-0.71

XL

-0.70

ISO

-0.67

itialized

-0.62

 Clown

-0.60

wagen

-0.60

xtap

-0.58

STD

-0.58

ESCO

-0.58

nikov

-0.57

POSITIVE LOGITS

 than

0.62

 Â»

0.56

 mins

0.56

 âĨĴ

0.55

 Juliet

0.53

 auth

0.52

 embed

0.51

 conservative

0.50

maker

0.50

 âĢº

0.50

Activations Density 0.029%

phrases related to news articles and journalism

No Comments

No Known Activations