INDEX

Explanations

information related to historical events or facts

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

-0.86

Ds

-0.80

heid

-0.70

Transfer

-0.69

 Skies

-0.64

ulner

-0.63

adelphia

-0.62

ourses

-0.62

SEA

-0.61

Nation

-0.61

POSITIVE LOGITS

 bitters

0.79

 forgiving

0.78

 backstory

0.74

 overboard

0.71

 than

0.69

 mortar

0.69

fetched

0.69

 luck

0.68

 cushion

0.67

 chew

0.67

Activations Density 0.152%

information related to historical events or facts

No Comments

No Known Activations