INDEX

Explanations

terms related to coding errors or vulnerabilities, specifically concerning buffer overflows and unsafe data handling

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

injection, overflow, attack, detected

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

appartamento

-0.38

hearsed

-0.37

cluster

-0.36

story

-0.36

 evenly

-0.36

vode

-0.35

 fairer

-0.35

Brainz

-0.35

 verbonden

-0.35

 story

-0.34

POSITIVE LOGITS

Искәрмәләр

0.53

CWE

0.50

 становника

0.49

 doInBackground

0.47

breos

0.47

onnaissance

0.46

VolleyError

0.45

FTFY

0.45

ffindor

0.45

PerformLayout

0.44

Activations Density 0.031%

terms related to coding errors or vulnerabilities, specifically concerning buffer overflows and unsafe data handling

injection, overflow, attack, detected

No Comments

No Known Activations