INDEX

Explanations

phrases related to adverse behavior or incidents

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Configuration

jbloom/Gemma-2b-IT-Residual-Stream-SAEs/gemma_2b_it_blocks.12.hook_resid_post_16384

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

HuggingFaceFW/fineweb

Features

16,384

Data Type

float32

Hook Name

blocks.12.hook_resid_post

Hook Layer

Architecture

standard

Context Size

1,024

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Mlle

-1.42

 Abbé

-1.39

Juf

-1.37

 Souha

-1.35

 reluct

-1.35

 indestru

-1.32

 Áng

-1.29

 unspeak

-1.29

 inconce

-1.28

 shenan

-1.28

POSITIVE LOGITS

EndInit

0.79

 NSCoder

0.68

GraphicsUnit

0.67

BeginInit

0.59

CodedInputStream

0.59

商品説明

0.58

Espèce

0.56

consulté

0.56

isContained

0.56

 ComVisible

0.55

Activations Density 0.072%

phrases related to adverse behavior or incidents

No Comments

No Known Activations