INDEX

Explanations

technical terms related to various products and services, including names of items in categories like household cleaning items, softwares, and websites

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

 Subtle

-0.70

Ĥª

-0.68

âĶģ

-0.63

 stoked

-0.61

humans

-0.61

ãĤ®

-0.60

nels

-0.60

kward

-0.59

lished

-0.58

BLIC

-0.57

POSITIVE LOGITS

ongevity

1.17

oyd

1.15

ateral

1.15

uggage

1.13

yrics

1.12

ibrarian

1.10

ifestyle

1.09

ayers

1.09

ibrary

1.08

opez

1.07

Activations Density 3.943%

technical terms related to various products and services, including names of items in categories like household cleaning items, softwares, and websites

No Comments

No Known Activations