INDEX

Explanations

locations

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

 respons

-0.72

onest

-0.69

 thereafter

-0.68

 thereof

-0.68

xxxx

-0.67

 encour

-0.67

 afterwards

-0.65

akespe

-0.64

 versa

-0.64

_.

-0.63

POSITIVE LOGITS

 Expand

0.82

][

0.81

reetings

0.76

 Introduction

0.75

 Vegan

0.75

 Updated

0.74

 Finder

0.74

zbollah

0.74

 Transcript

0.73

 Hearthstone

0.73

Activations Density 1.285%

locations

No Comments

No Known Activations