INDEX

Explanations

names of locations or political figures

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

TAIN

-0.85

CLASSIFIED

-0.71

quo

-0.70

ÄŁ

-0.68

cules

-0.67

cles

-0.62

 punishable

-0.62

ITAL

-0.61

 Phant

-0.58

 Intelligent

-0.58

POSITIVE LOGITS

endale

1.24

uten

1.14

itz

1.11

aston

1.10

acial

1.09

itch

1.08

ossip

1.07

itches

1.07

izzard

1.06

oster

1.05

Activations Density 0.019%

names of locations or political figures

No Comments

No Known Activations