INDEX

Explanations

political and governmental terms and names

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

zsche

-0.74

opter

-0.67

 Warfare

-0.66

origin

-0.66

AF

-0.61

Ø±

-0.60

rocket

-0.60

zac

-0.60

olog

-0.60

AAF

-0.60

POSITIVE LOGITS

 Governor

0.84

 vetoed

0.83

alty

0.82

 governor

0.80

Govern

0.80

elect

0.76

unicip

0.76

 mansion

0.75

atures

0.74

 gubernatorial

0.73

Activations Density 0.715%

political and governmental terms and names

No Comments

No Known Activations