INDEX

Explanations

comparisons between different groups or categories

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĤ¡

-0.72

 shenan

-0.68

iHUD

-0.68

oola

-0.64

 Emblem

-0.64

 Canaver

-0.64

ISM

-0.63

ocalypse

-0.62

ERN

-0.60

soType

-0.59

POSITIVE LOGITS

 peers

0.99

 counterparts

0.99

average

0.98

 average

0.90

 others

0.88

 nons

0.84

 those

0.84

 untreated

0.81

 comparable

0.80

normal

0.80

Activations Density 0.144%

comparisons between different groups or categories

No Comments

No Known Activations