INDEX

Explanations

references to rules and conditions associated with actions, particularly in the context of agreements or transactions

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

followed by adverbs or conjunctions

np_acts-logits-general · gemini-2.0-flash

actions taken unnecessarily

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-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 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

Према

-0.51

bably

-0.47

tioned

-0.46

ulton

-0.46

Становништво

-0.45

 تقری

-0.45

=?

-0.44

 remarkably

-0.44

toHexString

-0.44

subsec

-0.44

POSITIVE LOGITS

 unnecessarily

1.33

 needlessly

1.19

 unduly

1.00

 inneces

0.98

 prematurely

0.97

 anymore

0.97

 blindly

0.92

 willy

0.88

 unless

0.84

 terlalu

0.82

Activations Density 0.514%

references to rules and conditions associated with actions, particularly in the context of agreements or transactions

followed by adverbs or conjunctions

actions taken unnecessarily

No Comments

No Known Activations