INDEX

Explanations

references to procedural steps and instructions

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

step by step summary

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

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Taktlose

-0.49

Bronnen

-0.47

 medzi

-0.47

astify

-0.47

 blandt

-0.46

contentValues

-0.45

pulumi

-0.45

 Scalars

-0.45

 metálica

-0.44

Ecotoxicity

-0.44

POSITIVE LOGITS

 steps

0.67

step

0.67

 step

0.66

 detailed

0.61

 stepwise

0.59

steps

0.58

detailed

0.56

Steps

0.54

 Step

0.54

 Steps

0.54

Activations Density 0.006%

references to procedural steps and instructions

step by step summary

No Comments

No Known Activations