INDEX

Explanations

terms related to hypothetical scenarios, consequences, and possibilities

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Mant

-0.67

 Federation

-0.63

Moz

-0.62

 Lauder

-0.60

 Yards

-0.60

 Rising

-0.60

 Seeking

-0.59

 rejection

-0.59

FW

-0.58

 honoring

-0.58

POSITIVE LOGITS

't

1.77

adian

1.25

berra

1.24

NOT

1.23

 afford

1.23

 easily

1.17

isters

1.04

 safely

1.01

 manipulate

0.96

ister

0.95

Activations Density 1.746%

terms related to hypothetical scenarios, consequences, and possibilities

No Comments

No Known Activations