INDEX

Explanations

references to historical events and storytelling

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

 girls

-0.75

 breasts

-0.75

 boobs

-0.74

 womb

-0.72

 females

-0.71

 sisters

-0.71

 moms

-0.70

phrine

-0.70

 panties

-0.70

 mothers

-0.69

POSITIVE LOGITS

HAEL

0.97

 himself

0.93

 businessman

0.90

 Theodore

0.90

 preacher

0.90

 Rodrigo

0.89

Mr

0.87

 Dalai

0.87

 rapper

0.86

Mr

0.85

Activations Density 0.580%

references to historical events and storytelling

No Comments

No Known Activations