INDEX

Explanations

statements or actions related to public figures or political 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

 Gems

-0.66

 Mile

-0.64

classic

-0.64

services

-0.64

 pregn

-0.63

 Veter

-0.62

 Voltage

-0.62

 Claire

-0.62

 Mandatory

-0.60

 Impact

-0.58

POSITIVE LOGITS

zbollah

1.38

'll

1.30

'd

1.29

resy

1.22

eded

1.11

uristic

1.04

aven

1.03

ather

1.02

ctic

1.02

eding

1.01

Activations Density 2.402%

statements or actions related to public figures or political events

No Comments

No Known Activations