INDEX

Explanations

information related to monitoring and following activities, specifically in the context of technology advancements and business initiatives

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

DragonMagazine

-0.80

atana

-0.77

ramid

-0.74

çīĪ

-0.73

assi

-0.72

OLOGY

-0.72

dinand

-0.71

udos

-0.69

KEN

-0.69

haps

-0.69

POSITIVE LOGITS

 afloat

1.11

 tabs

1.08

 vigil

1.08

 pace

0.97

 vigilance

0.94

 track

0.89

 intact

0.89

 uninterrupted

0.85

stay

0.84

 quiet

0.82

Activations Density 3.732%

information related to monitoring and following activities, specifically in the context of technology advancements and business initiatives

No Comments

No Known Activations