INDEX

Explanations

instances where actions are taken or decisions are made

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

Negative Logits

 Regions

-0.66

 bills

-0.64

bots

-0.64

birds

-0.62

Kids

-0.62

Gov

-0.61

âĸº

-0.60

 Prompt

-0.60

byss

-0.60

YES

-0.60

POSITIVE LOGITS

ointed

1.23

 integral

1.04

agram

1.03

abolic

1.02

onym

1.02

oint

0.97

 invitation

0.97

 indication

0.97

 obstacle

0.97

omaly

0.96

Activations Density 0.122%

instances where actions are taken or decisions are made

No Known Activations