INDEX

Explanations

occurrences of numerical values or references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 more

-0.33

 players

-0.33

 generational

-0.32

 vitam

-0.32

 little

-0.32

 getItemId

-0.31

 zero

-0.30

 bomb

-0.30

 sequential

-0.30

 richer

-0.29

POSITIVE LOGITS

fromnode

0.65

KommentareTeilen

0.65

postsleuth

0.64

principalColumn

0.63

 ویکی‌پدیای

0.60

transQ

0.60

OCCURRED

0.59

enumi

0.59

 ویکی‌پدی

0.59

niſſe

0.59

Activations Density 0.019%

occurrences of numerical values or references

No Comments

No Known Activations