INDEX

Explanations

references to parts of a series such as book volumes, episodes, and chapters

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

death

-0.71

inventoryQuantity

-0.70

uts

-0.70

"))

-0.68

NK

-0.64

iverse

-0.64

onz

-0.62

doesn

-0.61

]"

-0.61

gif

-0.61

POSITIVE LOGITS

 being

1.07

 topping

1.03

 preferring

0.96

 needing

0.96

 appearing

0.95

 intervening

0.91

 boasting

0.90

 becoming

0.90

 agreeing

0.90

 ringing

0.89

Activations Density 0.401%

references to parts of a series such as book volumes, episodes, and chapters

No Comments

No Known Activations