INDEX

Explanations

times and sequences

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

 Amendments

-0.54

Beg

-0.53

 chatter

-0.53

CoC

-0.52

 Cere

-0.51

 Paste

-0.51

idious

-0.50

oidal

-0.50

 Emergency

-0.48

 Nurs

-0.47

POSITIVE LOGITS

etheless

0.97

©¶æ

0.94

Ĥİ

0.87

ħĭ

0.86

ilaterally

0.85

soever

0.82

til

0.81

¥ŀ

0.81

terday

0.81

ÃĥÃĤÃĥÃĤ

0.80

Activations Density 6.869%

times and sequences

No Comments

No Known Activations