INDEX

Explanations

technical and statistical terms related to documents and articles

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

Ã¨re

-0.83

ternity

-0.81

ooting

-0.78

ieving

-0.77

ses

-0.76

hooting

-0.76

omet

-0.76

nery

-0.75

overe

-0.74

mel

-0.74

POSITIVE LOGITS

tons

1.03

redients

0.89

 spree

0.78

 ability

0.75

AME

0.74

 mechanism

0.74

 heights

0.73

 agent

0.73

uez

0.72

 subsequ

0.71

Activations Density 1.244%

technical and statistical terms related to documents and articles

No Comments

No Known Activations