INDEX

Explanations

instances of numbers and their structures in a given dataset

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

This looks like low-level machine code or memory/byte-level data structures, not natural language tokens. Without more context, I cannot confidently characterize what this neuron is doing.

oai_token-act-pair · claude-3-5-haiku-20241022 Triggered by @neilrathi

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.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

 мәкал

-0.52

 Мексичка

-0.46

ⓧ

-0.46

 Autorizaciones

-0.46

 mergeFrom

-0.45

 bowiem

-0.44

]=>

-0.44

awtextra

-0.43

Obrázky

-0.41

帖最后由

-0.41

POSITIVE LOGITS

 assister

0.45

 assist

0.45

the

0.44

 drop

0.42

+#+#

0.41

 Assy

0.41

NewReader

0.41

 aggressive

0.40

 conf

0.40

cog

0.40

Activations Density 0.466%

instances of numbers and their structures in a given dataset

This looks like low-level machine code or memory/byte-level data structures, not natural language tokens. Without more context, I cannot confidently characterize what this neuron is doing.

No Comments

No Known Activations