INDEX

Explanations

dates expressed in years

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

 Shutterstock

-0.68

 Secrets

-0.65

 Advice

-0.65

 Purs

-0.65

 Challenges

-0.64

 Tags

-0.63

 Cambodia

-0.62

 Dreams

-0.61

ĺħ

-0.58

 Tools

-0.58

POSITIVE LOGITS

1.10

ocratic

0.96

âĢĳ

0.93

âĢĲ

0.90

-$

0.87

ennial

0.85

erous

0.83

%-

0.82

-[

0.81

-)

0.81

Activations Density 5.353%

dates expressed in years

No Comments

No Known Activations