INDEX

Explanations

terms related to the concept of "tuning."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 3-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_3/width_16k/average_l0_90

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.3.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

nos

-0.42

 evenings

-0.40

InsertCommand

-0.39

ായി

-0.38

 noDo

-0.38

 чемпиона

-0.37

 Saturdays

-0.37

SafeAreaView

-0.37

apatalk

-0.36

 mornings

-0.36

POSITIVE LOGITS

 tune

1.48

 Tune

1.33

tune

1.27

Tune

1.23

 tuning

1.13

 tuned

1.12

 tunnel

1.06

 Tuning

1.06

 Tunnel

1.04

Tun

1.03

Activations Density 0.094%

terms related to the concept of "tuning."

No Comments

No Known Activations