INDEX

Explanations

references to the author Gordon R. Dickson

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

Gordon + associated names

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

訟

-0.56

 évit

-0.55

scher

-0.55

Vict

-0.53

++];

-0.52

gaard

-0.52

 honte

-0.52

 bewer

-0.51

нутри

-0.51

punkt

-0.51

POSITIVE LOGITS

 Highlander

0.86

StructEnd

0.84

kháu

0.74

 Highlanders

0.70

 Sond

0.69

 "..\..\

0.66

izr

0.65

 Ramsay

0.65

 Mako

0.62

 Riproduzione

0.60

Activations Density 0.032%

references to the author Gordon R. Dickson

Gordon + associated names

No Comments

No Known Activations