INDEX

Explanations

numeric values, particularly numbers followed by specific context indicating dates or references to publications

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

numerical data or statistical references

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

months and years

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

tagHelper

-0.58

 OrderedDict

-0.56

 Souver

-0.54

soever

-0.49

 asupra

-0.49

stonia

-0.49

Scy

-0.48

economy

-0.47

 Moussa

-0.47

aimerais

-0.47

POSITIVE LOGITS

Dec

0.67

Nov

0.66

Oct

0.62

 estekak

0.61

Aug

0.61

Sep

0.58

Apr

0.58

eload

0.58

Feb

0.57

Jan

0.56

Activations Density 0.019%

numeric values, particularly numbers followed by specific context indicating dates or references to publications

numerical data or statistical references

months and years

No Comments

No Known Activations