INDEX

Explanations

mentions of the word "Gonz" or similar variations, potentially related to a specific person or entity

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Lauder

-0.71

voy

-0.69

 SPORTS

-0.68

âĢ¢âĢ¢

-0.68

croft

-0.61

 Wond

-0.61

 Skull

-0.60

 DEBUG

-0.60

fitting

-0.60

rek

-0.58

POSITIVE LOGITS

alez

1.38

etta

1.03

alo

1.01

onz

0.95

aga

0.88

osa

0.87

aido

0.83

alian

0.81

ole

0.77

onne

0.77

Activations Density 0.018%

mentions of the word "Gonz" or similar variations, potentially related to a specific person or entity

No Comments

No Known Activations