INDEX

Explanations

structured programming constructs and comments in code

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

TagMode

-0.90

expandindo

-0.85

 ProtoMessage

-0.83

 Administrativna

-0.83

OGND

-0.81

 ſind

-0.79

<unused41>

-0.75

<unused51>

-0.75

<unused8>

-0.75

[@BOS@]

-0.74

POSITIVE LOGITS

↵

0.35

 Decke

0.34

 geladeira

0.33

 either

0.33

0.32

La

0.32

 once

0.32

La

0.32

In

0.32

Activations Density 0.009%

structured programming constructs and comments in code

No Comments

No Known Activations