INDEX

Explanations

years expressed as numerical ranges

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

aila

-0.65

earable

-0.63

ellectual

-0.62

rim

-0.62

OTE

-0.62

 rall

-0.60

 Cosponsors

-0.59

ocus

-0.58

ISA

-0.58

estic

-0.57

POSITIVE LOGITS

1.03

0.98

0.96

 onwards

0.93

0.91

ruary

0.90

0.87

0.85

0.84

Activations Density 3.721%

years expressed as numerical ranges

No Comments

No Known Activations