INDEX

Explanations

references to events or things related to the past

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

utterstock

-0.89

vous

-0.86

Executive

-0.81

gur

-0.80

OTUS

-0.75

otiation

-0.75

rawdownloadcloneembedreportprint

-0.73

plementation

-0.72

rencies

-0.72

atography

-0.72

POSITIVE LOGITS

 fashioned

1.55

fashioned

1.03

 standby

1.02

 timers

1.00

school

0.87

timer

0.86

 Norse

0.81

owan

0.81

rums

0.80

 relic

0.79

Activations Density 1.428%

references to events or things related to the past

No Comments

No Known Activations