INDEX

Explanations

references to must-see or must-read content

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Slide

-0.75

 Reloaded

-0.75

 Bohem

-0.73

geist

-0.73

 Tempest

-0.70

 Feld

-0.69

 Aurora

-0.69

 Rounds

-0.68

 Bomb

-0.68

Ips

-0.68

POSITIVE LOGITS

dos

0.97

issue

0.94

answer

0.93

reads

0.91

contact

0.90

know

0.88

recogn

0.86

learn

0.86

carry

0.86

have

0.86

Activations Density 0.026%

references to must-see or must-read content

No Comments

No Known Activations