INDEX

Explanations

references to specific brands or products

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

Negative Logits

 Debor

-0.75

henko

-0.66

 proble

-0.65

Oro

-0.65

Alv

-0.63

 Stras

-0.63

aldi

-0.62

riot

-0.60

rov

-0.59

birth

-0.59

POSITIVE LOGITS

igans

1.13

ittle

1.10

igan

1.01

isted

0.93

ounge

0.90

ition

0.90

ingen

0.87

stra

0.85

mouth

0.84

abor

0.84

Activations Density 0.024%

references to specific brands or products

No Known Activations