INDEX

Explanations

words related to specific smells, particularly negative or unusual ones, and toxic substances

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

iscopal

-0.75

lisher

-0.75

umbn

-0.73

 pivot

-0.72

odder

-0.69

 Springer

-0.69

udeau

-0.66

ades

-0.66

 PowerPoint

-0.66

doms

-0.63

POSITIVE LOGITS

 smelling

1.09

 salts

1.00

 smells

0.96

 smell

0.95

 odor

0.94

 smelled

0.94

 faintly

0.93

 perfume

0.91

 emanating

0.89

 aroma

0.89

Activations Density 0.032%

words related to specific smells, particularly negative or unusual ones, and toxic substances

No Comments

No Known Activations