INDEX

Explanations

evaluations of experiences or objects ranging from best to worst

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

Superlative adjectives followed by "things"

np_acts-logits-general · gemini-2.0-flash

superlative evaluations and ever

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

 oczywiście

-0.59

 bezeichneter

-0.57

 становника

-0.56

}{*}{}

-0.54

MigrationBuilder

-0.54

/>";

-0.51

]-->

-0.51

 tfsi

-0.50

 виправивши

-0.49

***********/

-0.49

POSITIVE LOGITS

 ever

1.04

 greatest

0.93

 EVER

0.81

ever

0.80

greatest

0.80

 worst

0.79

 Ever

0.79

 strangest

0.78

Ever

0.76

EVER

0.76

Activations Density 0.106%

evaluations of experiences or objects ranging from best to worst

Superlative adjectives followed by "things"

superlative evaluations and ever

No Comments

No Known Activations