INDEX

Explanations

references to news articles or controversial discussions

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

 vulner

-0.81

 myster

-0.80

 mathemat

-0.79

 sacrific

-0.75

 limb

-0.75

 incorpor

-0.75

 charism

-0.73

 writ

-0.72

 trainers

-0.71

 condem

-0.70

POSITIVE LOGITS

ï¸ı

1.44

ï¸

1.02

vernment

0.99

âĹ¼

0.96

lean

0.95

ãĥĥãĥī

0.92

log

0.92

MQ

0.90

SpaceEngineers

0.90

deg

0.90

Activations Density 0.255%

references to news articles or controversial discussions

No Comments

No Known Activations