INDEX

Explanations

food-related descriptions and items

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

adra

-0.91

adan

-0.70

adier

-0.69

BIT

-0.69

è£ħ

-0.69

utra

-0.68

enos

-0.68

icipated

-0.67

eda

-0.67

owitz

-0.66

POSITIVE LOGITS

rive

0.92

BY

0.91

own

0.88

eals

0.86

river

0.83

irect

0.80

paren

0.80

dit

0.79

nesday

0.77

 aback

0.76

Activations Density 2.891%

food-related descriptions and items

No Comments

No Known Activations