jesuino

Traceback (most recent call last):
  File "/usr/local/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/usr/local/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/usr/local/lib/python3.10/site-packages/kfp/dsl/executor_main.py", line 109, in <module>
    executor_main()
  File "/usr/local/lib/python3.10/site-packages/kfp/dsl/executor_main.py", line 101, in executor_main
    output_file = executor.execute()
  File "/usr/local/lib/python3.10/site-packages/kfp/dsl/executor.py", line 415, in execute

jesuino

# PIPELINE DEFINITION
# Name: kale-pipeline
# Description: Train a Random Forest classifier on the Iris dataset
# Inputs:
#    max_depth: int [Default: 2.0]
#    n_estimators: int [Default: 500.0]
components:
  comp-evaluate-model-step:
    executorLabel: exec-evaluate-model-step
    inputDefinitions:

jesuino

apiVersion: v1
kind: Secret
metadata:
  name: devpi-secret
  namespace: kubeflow
type: Opaque
data:
  password: ''
--
apiVersion: v1

jesuino

$ kale --nb ./examples/base/candies_sharing.ipynb --kfp_host http://127.0.0.1:8080 --run_pipeline
2025-11-07 11:59:08 Kale podutils:255         [INFO]     Getting the base image of container...
2025-11-07 11:59:08 Kale kfutils:70           [INFO]     Retrieving PodDefaults applied to server...
2025-11-07 11:59:08 Kale kale                 [WARNING]  Could not retrieve PodDefaults. Reason: [Errno 2] No such file or directory: '/var/run/secrets/kubernetes.io/serviceaccount/namespace'
2025-11-07 11:59:08 Kale kale                 [INFO]     Compiling Pipeline into KFP DSL code
2025-11-07 11:59:08 Kale kale                 [INFO]     Saving generated code in /home/wsiqueir/projects/kale/.kale
2025-11-07 11:59:08 Kale kale                 [INFO]     Successfully saved generated code: /home/wsiqueir/projects/kale/.kale/kale-pipeline.kale.py
dsl_script_path: /home/wsiqueir/projects/kale/.kale/kale-pipeline.kale.py
/home/wsiqueir/projects/kale/.venv/lib/python3.10/site-packages/kfp/client/client.py:159: FutureWarning

jesuino

diff --git a/labextension/src/lib/CellUtils.ts b/labextension/src/lib/CellUtils.ts
index 2424b76..102b297 100644
--- a/labextension/src/lib/CellUtils.ts
+++ b/labextension/src/lib/CellUtils.ts
@@ -8,8 +8,6 @@ import {
 } from '@jupyterlab/nbformat';
 import { Notebook, NotebookActions, NotebookPanel } from '@jupyterlab/notebook';
 import { PartialJSONObject } from '@lumino/coreutils';
-// Project Components
-import NotebookUtilities from './NotebookUtils';

jesuino

diff --git a/labextension/src/widgets/LeftPanel.tsx b/labextension/src/widgets/LeftPanel.tsx
index 27e55be..fcf3145 100644
--- a/labextension/src/widgets/LeftPanel.tsx
+++ b/labextension/src/widgets/LeftPanel.tsx
@@ -34,6 +34,7 @@ import Commands from '../lib/Commands';
 import { PageConfig } from '@jupyterlab/coreutils';
 
 const KALE_NOTEBOOK_METADATA_KEY = 'kubeflow_notebook';
+const DEFAULT_EXPERIMENT_NAME = 'Default';
 

jesuino

diff --git a/labextension/src/components/Input.tsx b/labextension/src/components/Input.tsx
index 457b84d..044b0be 100644
--- a/labextension/src/components/Input.tsx
+++ b/labextension/src/components/Input.tsx
@@ -1,173 +1,3 @@
-// /*
-//  * Copyright 2020 The Kale Authors
-//  *
-//  * Licensed under the Apache License, Version 2.0 (the "License");
-//  * you may not use this file except in compliance with the License.

jesuino

2025-04-29 20:03:51,890 ERROR [io.qua.run.Quarkus] (Quarkus Main Thread) Error running Quarkus: java.lang.ExceptionInInitializerError
        at java.base/jdk.internal.misc.Unsafe.allocateInstance(Native Method)
        at java.base/java.lang.invoke.DirectMethodHandle.allocateInstance(DirectMethodHandle.java:506)
        at io.quarkus.runtime.Quarkus.run(Quarkus.java:76)
        at io.quarkus.runtime.Quarkus.run(Quarkus.java:48)
        at io.quarkus.runner.GeneratedMain.main(Unknown Source)
        at java.base/jdk.internal.reflect.DirectMethodHandleAccessor.invoke(DirectMethodHandleAccessor.java:104)
        at java.base/java.lang.reflect.Method.invoke(Method.java:565)
        at io.quarkus.runner.bootstrap.StartupActionImpl$1.run(StartupActionImpl.java:116)
        at java.base/java.lang.Thread.run(Thread.java:1447)

jesuino

package org.fxapps.ollamamcp;

import java.util.Map;

import io.modelcontextprotocol.client.McpClient;
import io.modelcontextprotocol.client.transport.ServerParameters;
import io.modelcontextprotocol.client.transport.StdioClientTransport;
import io.modelcontextprotocol.spec.McpSchema.CallToolRequest;
import io.modelcontextprotocol.spec.McpSchema.TextContent;

jesuino

Eu: então além da matriz com os embeddings de um token eu ainda preciso passar essa máscara de entrada para uma LLM? isso é comum em todos os modelos de LLM?

deepseek: Sim, além da matriz de embeddings dos tokens, você geralmente precisa passar uma máscara de atenção (attention mask) para o modelo. Isso é comum na maioria dos modelos de LLM modernos, especialmente aqueles baseados em arquiteturas Transformer, como BERT, GPT, T5, entre outros.

Por que a máscara de atenção é necessária?

1. Ignorar Tokens de Preenchimento:
   • A máscara de atenção informa ao modelo quais tokens são reais e quais são tokens de preenchimento (padding).
   • Isso é crucial porque os tokens de preenchimento não têm significado semântico e não devem influenciar o processamento do modelo.