INDEX

Explanations

references to software or technology platforms

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

otine

-0.70

ovie

-0.70

affe

-0.63

 Mile

-0.63

 Estate

-0.62

 Ruin

-0.61

 Mash

-0.60

 Disciple

-0.59

 Rite

-0.58

 Spartan

-0.57

POSITIVE LOGITS

fully

0.81

 everywhere

0.79

ifully

0.79

ally

0.77

 elsewhere

0.76

actly

0.75

 enough

0.73

ably

0.73

 comparable

0.72

worthy

0.72

Activations Density 0.362%

references to software or technology platforms

No Comments

No Known Activations