INDEX

Explanations

references to competitive sports teams and matches

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

 Donation

-0.81

 memos

-0.75

eret

-0.74

 memo

-0.73

Asset

-0.72

 condol

-0.70

 reinvest

-0.68

endum

-0.67

EngineDebug

-0.67

Product

-0.66

POSITIVE LOGITS

foe

1.27

 opponent

1.25

 foes

1.12

 opponents

1.08

 rivals

1.08

 adversaries

1.05

 enemy

1.04

 adversary

1.04

 rival

0.98

 challengers

0.97

Activations Density 0.276%

references to competitive sports teams and matches

No Comments

No Known Activations