INDEX

Explanations

references to publications and media outlets, such as newspapers, journals, magazines, and news agencies

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

ividual

-0.75

"]=>

-0.73

matter

-0.71

 exting

-0.70

ãĥĥãĥī

-0.70

ento

-0.69

Female

-0.68

ascript

-0.68

quist

-0.67

////////////////////////////////

-0.66

POSITIVE LOGITS

 Artemis

0.98

 Spart

0.93

 Ares

0.90

 Ceres

0.84

Tup

0.79

USS

0.78

 Amos

0.77

 Martha

0.77

HMS

0.77

Zac

0.75

Activations Density 1.418%

references to publications and media outlets, such as newspapers, journals, magazines, and news agencies

No Comments

No Known Activations