INDEX

Explanations

political figures and events

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

âĶĢâĶĢ

-0.90

ModLoader

-0.84

Â·Â·

-0.73

ãĥŁ

-0.70

LCS

-0.65

 railways

-0.63

 Pradesh

-0.62

FACE

-0.61

Pokemon

-0.61

 subsistence

-0.61

POSITIVE LOGITS

zinski

1.14

inski

1.09

kowski

1.06

isner

1.06

ovich

1.05

linger

1.05

antz

1.03

enson

1.02

aney

1.00

enberg

0.99

Activations Density 2.362%

political figures and events

No Comments

No Known Activations