INDEX

Explanations

sentences or quotations indicating strong praise or admiration

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

 conver

-0.71

 allied

-0.69

 isolate

-0.68

 decap

-0.67

 contrace

-0.67

aper

-0.66

 doomed

-0.66

 periodic

-0.65

apers

-0.65

 costly

-0.65

POSITIVE LOGITS

 âĢķ

1.02

["

0.90

>>\

0.86

[/

0.78

Written

0.77

isSpecialOrderable

0.73

 Translation

0.73

pg

0.71

BuyableInstoreAndOnline

0.69

/"

0.69

Activations Density 0.413%

sentences or quotations indicating strong praise or admiration

No Comments

No Known Activations