INDEX

Explanations

timestamps and related details in a blog post

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

 alike

-0.65

unts

-0.64

 Cantor

-0.62

oglu

-0.58

uits

-0.57

idia

-0.57

antis

-0.57

hesda

-0.56

 setups

-0.56

naires

-0.56

POSITIVE LOGITS

DragonMagazine

0.66

ãĥ¼ãĤ¯

0.63

Reader

0.62

ãĤ¼

0.62

ãĥ¼ãĥ³

0.60

channelAvailability

0.59

cair

0.59

 Duration

0.58

ãĥ´ãĤ¡

0.58

©¶æ¥µ

0.58

Activations Density 0.077%

timestamps and related details in a blog post

No Comments

No Known Activations