INDEX

Explanations

phrases indicating a contrast or surprising outcome

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

emia

-0.73

pour

-0.68

ahime

-0.67

 antioxid

-0.66

iant

-0.65

PI

-0.61

redients

-0.61

States

-0.61

åĤ

-0.59

iants

-0.59

POSITIVE LOGITS

theless

1.60

 dreamed

1.12

 imagined

0.87

 bothered

0.87

 existed

0.86

 heard

0.85

 doubted

0.85

lasting

0.78

entimes

0.75

more

0.75

Activations Density 0.958%

phrases indicating a contrast or surprising outcome

No Comments

No Known Activations