INDEX

Explanations

proper nouns, particularly names of authors or book titles

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

 كومونز

-0.75

Poznám

-0.69

pueden

-0.65

 ricev

-0.64

enzuela

-0.62

 ModelExpression

-0.60

 pinak

-0.59

después

-0.59

algunos

-0.57

 himo

-0.57

POSITIVE LOGITS

 subgoals

0.56

 inappro

0.51

 extré

0.51

 sokak

0.51

 célé

0.49

 desnuda

0.49

Jr

0.49

 Hitam

0.49

(@

0.48

 vecteur

0.48

Activations Density 0.308%

proper nouns, particularly names of authors or book titles

No Comments

No Known Activations