INDEX

Explanations

terms related to the color 'black'

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

PsyNetMessage

-0.91

ICLE

-0.77

mble

-0.75

ablishment

-0.75

igslist

-0.73

 Lauder

-0.72

agall

-0.71

OPLE

-0.71

=-=-=-=-

-0.70

xon

-0.69

POSITIVE LOGITS

smith

1.27

ened

1.19

berry

1.11

jack

1.09

listed

1.01

berries

0.98

hawk

0.96

bird

0.96

birds

0.92

beard

0.91

Activations Density 0.063%

terms related to the color 'black'

No Comments

No Known Activations