release 0.3

demo/demo3/paddlepalm

+../../paddlepalm/
\ No newline at end of file

demo/demo3/pretrain

+/home/zhangyiming/yiming/v02-exe/pretrain
\ No newline at end of file

demo/demo3/run.py

+# coding=utf-8
 import paddlepalm as palm
+import json
 
 if __name__ == '__main__':
 
     max_seqlen = 512
-    batch_size = 32
+    batch_size = 3
+    num_epochs = 2
+    lr = 1e-3
+    vocab_path = './pretrain/ernie/vocab.txt'
+
+    train_file = './data/cls4mrqa/train.tsv'
+
+    config = json.load(open('./pretrain/ernie/ernie_config.json'))
+    print(config)
+    ernie = palm.backbone.ERNIE.from_config(config)
+    pred_ernie = palm.backbone.ERNIE.from_config(config, phase='pred')
+
+    # cls_reader2 = palm.reader.cls(train_file_topic, vocab_path, batch_size, max_seqlen)
+    # cls_reader3 = palm.reader.cls(train_file_subj, vocab_path, batch_size, max_seqlen)
+    # topic_trainer = palm.Trainer('topic_cls', cls_reader2, cls)
+    # subj_trainer = palm.Trainer('subj_cls', cls_reader3, cls)
+
+    # 创建该分类任务的reader，由诸多参数控制数据集读入格式、文件数量、预处理规则等
+    cls_reader = palm.reader.ClassifyReader(vocab_path, batch_size, max_seqlen)
+    print(cls_reader.outputs_attr)
+    # 不同的backbone会对任务reader有不同的特征要求，例如对于分类任务，基本的输入feature为token_ids和label_ids，但是对于BERT，还要求从输入中额外提取position、segment、input_mask等特征，因此经过register后，reader会自动补充backbone所要求的字段
+    cls_reader.register_with(ernie)
+    print(cls_reader.outputs_attr)
+    # 创建任务头（task head），如分类、匹配、机器阅读理解等。每个任务头有跟该任务相关的必选/可选参数。注意，任务头与reader是解耦合的，只要任务头依赖的数据集侧的字段能被reader提供，那么就是合法的
+    cls_head = palm.head.Classify(4, 1024, 0.1)
+    cls_pred_head = palm.head.Classify(4, 1024, 0.1, phase='pred')
+
+    # 根据reader和任务头来创建一个训练器trainer，trainer代表了一个训练任务，内部维护着训练进程、和任务的关键信息，并完成合法性校验，该任务的模型保存、载入等相关规则控制
+    trainer = palm.Trainer('senti_cls', cls_reader, cls_head, save_predict_model=True, \
+                           pred_head=cls_pred_head, save_path='./output')
+
+    # match4mrqa.reuse_head_with(mrc4mrqa)
+
+    # data_vars = cls_reader.build()
+    # output_vars = ernie.build(data_vars)
+    # cls_head.build({'backbone': output_vars, 'reader': data_vars})
+
+    loss_var = trainer.build_forward(ernie, pred_ernie)
+
+    # controller.build_forward()
+    # Error! a head/backbone can be only build once! Try NOT to call build_forward method for any Trainer!
+
+    print(trainer.num_examples)
+    trainer.load_data(train_file, 'csv', num_epochs=2, batch_size=32)
+    print(trainer.num_examples)
+
+    n_steps = trainer.num_examples * num_epochs // batch_size
+    warmup_steps = int(0.1 * n_steps)
+    print(warmup_steps)
+    sched = palm.lr_sched.TriangularSchedualer(warmup_steps, n_steps)
+
+    adam = palm.optimizer.Adam(loss_var, lr, sched)
+
+    trainer.build_backward(optimizer=adam, weight_decay=0.001, \
+                           use_ema=True, ema_decay=0.999)
+
+    trainer.random_init_params()
+    trainer.load_pretrain('../../pretrain_model/ernie/params')
+
+    # trainer.train_one_step()
+    # trainer.train_one_epoch()
+    trainer.train()
+    trainer.save()
+
+
 
-    match_reader = palm.reader.match(train_file, vocab, \
-        max_seqlen, file_format='csv', tokenizer='wordpiece', \
-        lang='en', shuffle_train=True)
-    mrc_reader = palm.reader.mrc(train_file, phase='train')
-    mlm_reader = palm.reader.mlm(train_file, phase='train')
-    palm.reader.
 
-    match = palm.tasktype.cls(num_classes=4)
-    mrc = palm.tasktype.match(learning_strategy='pairwise')
-    mlm = palm.tasktype.mlm()
-    mlm.print()
 
-    
-    bb_flags = palm.load_json('./pretrain/ernie/ernie_config.json')
-    bb = palm.backbone.ernie(bb_flags['xx'], xxx)
-    bb.print()
 
-    match4mrqa = palm.Task('match4mrqa', match_reader, match_tt)
-    mrc4mrqa = palm.Task('match4mrqa', match_reader, match_tt)
 
-    # match4mrqa.reuse_with(mrc4mrqa)
 
 
-    controller = palm.Controller([mrqa, match4mrqa, mlm4mrqa])
+    # controller = palm.Controller([mrqa, match4mrqa, mlm4mrqa])
 
     loss = controller.build_forward(bb, mask_task=[])
 

paddlepalm/__init__.py

+import downloader
+from mtl_controller import Controller 
 
-import sys
-from paddlepalm.mtl_controller import Controller
-from paddlepalm.task_instance import Task
-
-sys.path.append('paddlepalm')
 
+del interface
+del task_instance
+del default_settings
+del utils
+del mtl_controller
\ No newline at end of file

paddlepalm/_downloader.py

+# -*- coding: UTF-8 -*-
+#   Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import print_function
+import os
+import tarfile
+import shutil
+try:
+    from urllib.request import urlopen # Python 3
+except ImportError:
+    from urllib2 import urlopen # Python 2
+
+
+import ssl
+
+__all__ = ["download", "ls"]
+
+# for https
+ssl._create_default_https_context = ssl._create_unverified_context
+
+_items = {
+    'pretrain': {'ernie-en-uncased-large': 'https://ernie.bj.bcebos.com/ERNIE_Large_en_stable-2.0.0.tar.gz',
+                 'bert-en-uncased-large': 'https://bert-models.bj.bcebos.com/uncased_L-24_H-1024_A-16.tar.gz',
+                 'bert-en-uncased-base': 'https://bert-models.bj.bcebos.com/uncased_L-12_H-768_A-12.tar.gz',
+                 'utils': None},
+    'reader': {'utils': None},
+    'backbone': {'utils': None},
+    'tasktype': {'utils': None},
+}
+
+def _download(item, scope, path, silent=False):
+    data_url = _items[item][scope]
+    if data_url == None:
+        return
+    if not silent:
+        print('Downloading {}: {} from {}...'.format(item, scope, data_url))
+    data_dir = path + '/' + item + '/' + scope
+    if not os.path.exists(data_dir):
+        os.makedirs(os.path.join(data_dir))
+    data_name = data_url.split('/')[-1]
+    filename = data_dir + '/' + data_name
+
+    # print process
+    def _chunk_report(bytes_so_far, total_size):
+        percent = float(bytes_so_far) / float(total_size)
+        if percent > 1:
+            percent = 1
+        if not silent:
+            print('\r>> Downloading... {:.1%}'.format(percent), end = "")
+    
+    # copy to local
+    def _chunk_read(response, url, chunk_size = 16 * 1024, report_hook = None):
+        total_size = response.info().getheader('Content-Length').strip()
+        total_size = int(total_size)
+        bytes_so_far = 0
+        with open("%s" % filename, "wb") as f:
+            while 1:
+                chunk = response.read(chunk_size)
+                f.write(chunk)
+                f.flush() 
+                bytes_so_far += len(chunk)
+                if not chunk:
+                    break
+                if report_hook:
+                    report_hook(bytes_so_far, total_size)
+        return bytes_so_far
+
+    response = urlopen(data_url)
+    _chunk_read(response, data_url, report_hook=_chunk_report)
+    
+    if not silent:
+        print(' done!')
+    
+    if item == 'pretrain':
+        if not silent:
+            print ('Extracting {}...'.format(data_name), end=" ")
+        if os.path.exists(filename):
+            tar = tarfile.open(filename, 'r')
+            tar.extractall(path = data_dir)
+            tar.close()
+            os.remove(filename)
+        if scope.startswith('bert'):
+            source_path = data_dir + '/' + data_name.split('.')[0]
+            fileList = os.listdir(source_path)
+            for file in fileList:
+                filePath = os.path.join(source_path, file)
+                shutil.move(filePath, data_dir)
+            os.removedirs(source_path)
+        if not silent:
+            print ('done!')
+        if not silent:
+            print ('Converting params...', end=" ")
+        _convert(data_dir, silent)
+        if not silent:
+            print ('done!')
+
+
+def _convert(path, silent=False):
+    if os.path.isfile(path + '/params/__palminfo__'):
+        if not silent:
+            print ('already converted.')
+    else:
+        if os.path.exists(path + '/params/'):
+            os.rename(path + '/params/', path + '/params1/')
+            os.mkdir(path + '/params/')
+            tar_model = tarfile.open(path + '/params/' + '__palmmodel__', 'w')
+            tar_info = open(path + '/params/'+ '__palminfo__', 'w')
+            for root, dirs, files in os.walk(path + '/params1/'):
+                for file in files:
+                    src_file = os.path.join(root, file)
+                    tar_model.add(src_file, '__paddlepalm_' + file)
+                    tar_info.write('__paddlepalm_' + file)
+                    os.remove(src_file)
+            tar_model.close()
+            tar_info.close()
+            os.removedirs(path + '/params1/') 
+
+def download(item, scope='all', path='.'):
+    item = item.lower()
+    scope = scope.lower()
+    assert item in _items, '{} is not found. Support list: {}'.format(item, list(_items.keys()))
+   
+    if _items[item]['utils'] is not None:
+        _download(item, 'utils', path, silent=True)
+
+    if scope != 'all':
+        assert scope in _items[item], '{} is not found. Support scopes: {}'.format(scope, list(_items[item].keys()))
+        _download(item, scope, path)
+    else:
+        for s in _items[item].keys():
+            _download(item, s, path)
+
+
+def _ls(item, scope, l = 10):
+    if scope != 'all':
+        assert scope in _items[item], '{} is not found. Support scopes: {}'.format(scope, list(_items[item].keys()))
+        print ('{}'.format(scope))
+    else:
+        for s in _items[item].keys():
+            if s == 'utils':
+                continue
+            print ('  => '+s)
+
+def ls(item='all', scope='all'):
+    
+    if scope == 'utils':
+        return
+    if item != 'all':
+        assert item in _items, '{} is not found. Support scopes: {}'.format(item, list(_items.keys()))
+        print ('Available {} items:'.format(item))
+        _ls(item, scope)
+    else:
+        l = max(map(len, _items.keys()))
+        for i in _items.keys():
+            print ('Available {} items: '.format(i))
+            _ls(i, scope, l)
+
+
+    

paddlepalm/backbone/bert.py

@@ -52,9 +52,9 @@ class Model(backbone):
 
     @property
     def inputs_attr(self):
-        return {"token_ids": [[-1, -1, 1], 'int64'],
-                "position_ids": [[-1, -1, 1], 'int64'],
-                "segment_ids": [[-1, -1, 1], 'int64'],
+        return {"token_ids": [[-1, -1], 'int64'],
+                "position_ids": [[-1, -1], 'int64'],
+                "segment_ids": [[-1, -1], 'int64'],
                 "input_mask": [[-1, -1, 1], 'float32']}
 
     @property
@@ -73,7 +73,7 @@ class Model(backbone):
 
         self._emb_dtype = 'float32'
         # padding id in vocabulary must be set to 0
-        emb_out = fluid.layers.embedding(
+        emb_out = fluid.embedding(
             input=src_ids,
             size=[self._voc_size, self._emb_size],
             dtype=self._emb_dtype,
@@ -84,14 +84,14 @@ class Model(backbone):
         # fluid.global_scope().find_var('backbone-word_embedding').get_tensor()
         embedding_table = fluid.default_main_program().global_block().var(scope_name+self._word_emb_name)
         
-        position_emb_out = fluid.layers.embedding(
+        position_emb_out = fluid.embedding(
             input=pos_ids,
             size=[self._max_position_seq_len, self._emb_size],
             dtype=self._emb_dtype,
             param_attr=fluid.ParamAttr(
                 name=scope_name+self._pos_emb_name, initializer=self._param_initializer))
 
-        sent_emb_out = fluid.layers.embedding(
+        sent_emb_out = fluid.embedding(
             sent_ids,
             size=[self._sent_types, self._emb_size],
             dtype=self._emb_dtype,

paddlepalm/backbone/ernie.py

@@ -62,11 +62,11 @@ class Model(backbone):
 
     @property
     def inputs_attr(self):
-        return {"token_ids": [[-1, -1, 1], 'int64'],
-                "position_ids": [[-1, -1, 1], 'int64'],
-                "segment_ids": [[-1, -1, 1], 'int64'],
+        return {"token_ids": [[-1, -1], 'int64'],
+                "position_ids": [[-1, -1], 'int64'],
+                "segment_ids": [[-1, -1], 'int64'],
                 "input_mask": [[-1, -1, 1], 'float32'],
-                "task_ids": [[-1,-1, 1], 'int64']}
+                "task_ids": [[-1,-1], 'int64']}
 
     @property
     def outputs_attr(self):
@@ -85,7 +85,7 @@ class Model(backbone):
         task_ids = inputs['task_ids']
 
         # padding id in vocabulary must be set to 0
-        emb_out = fluid.layers.embedding(
+        emb_out = fluid.embedding(
             input=src_ids,
             size=[self._voc_size, self._emb_size],
             dtype=self._emb_dtype,
@@ -96,14 +96,14 @@ class Model(backbone):
         # fluid.global_scope().find_var('backbone-word_embedding').get_tensor()
         embedding_table = fluid.default_main_program().global_block().var(scope_name+self._word_emb_name)
         
-        position_emb_out = fluid.layers.embedding(
+        position_emb_out = fluid.embedding(
             input=pos_ids,
             size=[self._max_position_seq_len, self._emb_size],
             dtype=self._emb_dtype,
             param_attr=fluid.ParamAttr(
                 name=scope_name+self._pos_emb_name, initializer=self._param_initializer))
 
-        sent_emb_out = fluid.layers.embedding(
+        sent_emb_out = fluid.embedding(
             sent_ids,
             size=[self._sent_types, self._emb_size],
             dtype=self._emb_dtype,
@@ -113,7 +113,7 @@ class Model(backbone):
         emb_out = emb_out + position_emb_out
         emb_out = emb_out + sent_emb_out
 
-        task_emb_out = fluid.layers.embedding(
+        task_emb_out = fluid.embedding(
             task_ids,
             size=[self._task_types, self._emb_size],
             dtype=self._emb_dtype,

paddlepalm/downloader.py

-
-from __future__ import print_function
-import os
-
-items = {
-    'pretrain': {'ernie-en-uncased-large': 'http://xxxxx',
-                 'xxx': 'xxx',
-                 'utils': None}
-    'reader': {'cls': 'xxx',
-               'xxx': 'xxx',
-               'utils': 'xxx'}
-    'backbone': {xxx}
-    'tasktype': {xxx}
-}
-
-
-def download(item, scope='all', path='.'):
-    item = item.lower()
-    scope = scope.lower()
-    assert item in items, '{} is not found. Support list: {}'.format(item, list(items.keys()))
-
-    if not os.path.exists(path, item):
-        os.makedirs(os.path.join(path, item))
-
-    def _download(item, scope, silent=False):
-        if not silent:
-            print('downloading {}: {} from {}...'.format(item, scope, items[item][scope]), end='')
-        urllib.downloadxxx(items[item][scope])
-        if not silent:
-            print('done!')
-        
-    if items['utils'] is not None:
-        _download(item, 'utils', silent=True)
-
-    if scope != 'all':
-        assert scope in items[item], '{} is not found. Support scopes: {}'.format(item, list(items[item].keys()))
-        _download(item, scope)
-    else:
-        for s in items[item].keys():
-            _download(item, s)
-
-
-def ls(item=None, scope='all'):
-    pass
-
+from _downloader import *
 

paddlepalm/mtl_controller.py

+# -*- coding: UTF-8 -*-
+#   Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import print_function
+
+import os
+import sys
+import importlib
+import multiprocessing
+from paddle import fluid
+from paddle.fluid import layers
+import yaml
+import json
+import logging
+import time
+import numpy as np
+
+from paddlepalm.utils.saver import init_pretraining_params, init_checkpoint
+from paddlepalm.utils.config_helper import PDConfig
+from paddlepalm.utils.print_helper import print_dict
+from paddlepalm.utils.reader_helper import create_net_inputs, create_iterator_fn, create_joint_iterator_fn, merge_input_attrs 
+
+from paddlepalm.default_settings import *
+from task_instance import TaskInstance, check_instances
+
+import Queue
+from threading import Thread
+
+DEBUG=False
+VERBOSE=0
+
+def _get_basename(f):
+    return os.path.splitext(f)[0]
+
+
+def _get_suffix(f):
+    return os.path.splitext(f)[-1]
+
+
+def _parse_yaml(f, asdict=True, support_cmd_line=False):
+    assert os.path.exists(f), "file {} not found.".format(f)
+    if support_cmd_line:
+        args = PDConfig(yaml_file=f, fuse_args=True)
+        args.build()
+        return args.asdict() if asdict else args
+    else:
+        if asdict:
+            with open(f, "r") as fin: 
+                yaml_config = yaml.load(fin, Loader=yaml.SafeLoader)
+            return yaml_config
+        else:
+            raise NotImplementedError()
+
+
+def _parse_json(f, asdict=True, support_cmd_line=False):
+    assert os.path.exists(f), "file {} not found.".format(f)
+    if support_cmd_line:
+        args = PDConfig(json_file=f, fuse_args=support_cmd_line)
+        args.build()
+        return args.asdict() if asdict else args
+    else:
+        if asdict:
+            with open(f, "r") as fin: 
+                config = json.load(fin)
+            return config
+        else:
+            raise NotImplementedError()
+            
+
+def _parse_list(string, astype=str):
+    assert isinstance(string, str), "{} is not a string.".format(string)
+    if ',' not in string:
+        return [astype(string)]
+    string = string.replace(',', ' ')
+    return [astype(i) for i in string.split()]
+
+
+def _try_float(s):
+    try:
+        float(s)
+        return(float(s))
+    except:
+        return s
+
+
+def _check_conf(conf, checklist=None):
+    assert isinstance(conf, dict), "{} is not a dict.".format(conf)
+    ret = {}
+    for k,v in conf.items():
+        if isinstance(v, str):
+            v = _try_float(v)
+        ret[k] = v
+    if checklist is not None:
+        for k, t in checklist:
+            assert k in ret, "required argument {} is NOT exist in config file.".format(k)
+            assert isintance(ret[k], t), "value type of argument {} should be {}".format(k, t)
+    return ret
+
+
+# TODO: 增加None机制，允许hidden size、batch size和seqlen设置为None
+def _check_io(in_attr, out_attr, strict=False, in_name="left", out_name="right"):
+    for name, attr in in_attr.items():
+        assert name in out_attr, in_name+': '+name+' not found in '+out_name
+        if attr != out_attr[name]:
+            if strict:
+                raise ValueError(name+': shape or dtype not consistent!')
+            else:
+                logging.warning('{}: shape or dtype not consistent!\n{}:\n{}\n{}:\n{}'.format(name, in_name, attr, out_name, out_attr[name]))
+
+
+def _merge_conf(conf1, conf2, conf1_first=True, strict=False):
+    assert isinstance(conf1, dict), "{} is not a dict.".format(conf1)
+    assert isinstance(conf2, dict), "{} is not a dict.".format(conf2)
+    base_conf = conf2 if conf1_first else conf1
+    base_conf = base_conf.copy()
+    new_conf = conf1 if conf1_first else conf2
+
+    for k, v in new_conf.items():
+        if k in base_conf:
+            if base_conf[k] != v:
+                raise Warning("value of argument {} has been updated to {}.".format(k, v))
+        else:
+            if strict:
+                continue
+            
+        base_conf[k] = v
+    return base_conf
+
+
+def _encode_inputs(inputs, scope_name, sep='/', cand_set=None):
+    outputs = {}
+    for k, v in inputs.items():
+        if cand_set is not None:
+            if k in cand_set:
+                outputs[k] = v
+            if scope_name+sep+k in cand_set:
+                outputs[scope_name+sep+k] = v
+        else:
+            outputs[scope_name+sep+k] = v
+    return outputs
+
+
+def _decode_inputs(inputs, scope_name, sep='/', keep_unk_keys=True):
+    outputs = {}
+    for name, value in inputs.items():
+        # var for backbone are also available to tasks
+        if keep_unk_keys and sep not in name:
+            outputs[name] = value
+        # var for this inst
+        if name.startswith(scope_name+'/'):
+            outputs[name[len(scope_name+'/'):]] = value
+    return outputs
+
+
+def _init_env(use_gpu):
+    if use_gpu:
+        place = fluid.CUDAPlace(0)
+        dev_count = fluid.core.get_cuda_device_count()
+    else:
+        place = fluid.CPUPlace()
+        dev_count = int(os.environ.get('CPU_NUM', multiprocessing.cpu_count()))
+    return fluid.Executor(place), dev_count
+
+
+def _fit_attr(conf, fit_attr, strict=False):
+    for i, attr in fit_attr.items():
+        if i not in conf:
+            if strict:
+                raise Exception('Argument {} is required to create a controller.'.format(i))
+            else:
+                continue
+        conf[i] = attr(conf[i])
+    return conf
+
+
+class Controller(object):
+
+    def __init__(self, config, task_dir='.', for_train=True):
+        """
+        Args:
+            config: (str|dict) 字符串类型时，给出yaml格式的config配置文件路径；
+        """
+
+        self._for_train = for_train
+        assert isinstance(config, str) or isinstance(config, dict), "a config dict or config file path is required to create a Controller."
+
+        if isinstance(config, str):
+            mtl_conf = _parse_yaml(config, support_cmd_line=True)
+        else:
+            mtl_conf = config
+                
+        mtl_conf = _check_conf(mtl_conf)
+        mtl_conf = _fit_attr(mtl_conf, REQUIRED_ARGS, strict=True)
+        mtl_conf = _fit_attr(mtl_conf, OPTIONAL_ARGS, strict=False)
+
+        exe, dev_count = _init_env(use_gpu=mtl_conf.get('use_gpu', True))
+        self.exe = exe
+        self.dev_count = dev_count
+
+        print_dict(mtl_conf, title='global configuration')
+
+        # parse task instances and target tags
+        instnames = _parse_list(mtl_conf['task_instance'])
+        assert len(instnames) == len(set(instnames)), "repeated task_instance is NOT supported."
+        num_instances = len(instnames)
+        self.num_instances = num_instances
+
+        instname_to_conf = {}
+        instname_to_id = {}
+        for id, instname in enumerate(instnames):
+            instpath = os.path.join(task_dir, instname+'.yaml')
+            conf = _parse_yaml(instpath, support_cmd_line=False)
+            # conf = _check_conf(conf, TASK_INSTANCE_REQUIRED_ARGS)
+            conf = _check_conf(conf)
+            temp_conf = _merge_conf(mtl_conf, conf, strict=True)
+            print_dict(temp_conf, title='{} configuration'.format(instname))
+            conf = _merge_conf(mtl_conf, conf)
+            
+            instname_to_conf[instname] = conf
+            instname_to_id[instname] = id
+
+        # prepare backbone
+        if 'backbone_config_path' in mtl_conf:
+            bb_conf = _parse_json(mtl_conf['backbone_config_path'])
+            bb_conf = _merge_conf(mtl_conf, bb_conf)
+        else:
+            bb_conf = mtl_conf
+        print_dict(bb_conf, title = 'backbone configuration'.format(instname))
+
+        bb_name = mtl_conf['backbone']
+        bb_mod = importlib.import_module(BACKBONE_DIR + '.' + bb_name)
+        Backbone = getattr(bb_mod, 'Model')
+
+        # create task instances
+        instances = []
+        for name in instnames:
+            instances.append(TaskInstance(name, instname_to_id[name], instname_to_conf[name]))
+
+        check_instances(instances)
+
+        # parse target_tag
+        if 'target_tag' in mtl_conf:
+            target_tag = str(mtl_conf['target_tag'])
+            tags = _parse_list(target_tag, astype=int)
+            assert len(tags) == len(instnames), "number of target_tag is NOT consistent with that in task_instance."
+            for tag, inst in zip(tags, instances):
+                inst.is_target = tag
+        else:
+            tags = [i.is_target for i in instances]
+        num_targets = sum(tags)
+        num_auxes = num_instances - num_targets
+
+        # parse mix ratios
+        if 'mix_ratio' in mtl_conf:
+            mix_ratio = str(mtl_conf['mix_ratio'])
+            mrs = _parse_list(mix_ratio, astype=float)
+            assert len(mrs) == num_instances, "number of mix_ratios is NOT consistent with num_instances."
+        else:
+            mrs = [1.0] * num_instances
+
+        for mr, inst in zip(mrs, instances):
+            inst.mix_ratio = mr
+
+        # parse task layer reuse tags
+        instname_to_reusehost = {i:i for i in instnames}
+        if 'task_reuse_tag' in mtl_conf:
+            tags = _parse_list(mtl_conf['task_reuse_tag'], astype=int)
+            assert len(tags) == num_targets, 'number of reuse_tags is NOT consistent with number of instances.'
+        else:
+            tags = []
+            mapper = {}
+            for inst in instances:
+                history = set()
+                history.add(inst.name)
+                cur_inst = inst
+                while True:
+                    if cur_inst.task_reuse_scope in history:
+                        mapper[inst.name] = len(tags)
+                        break
+                    elif cur_inst.task_reuse_scope in mapper:
+                        mapper[inst.name] = mapper[cur_inst.task_reuse_scope]
+                        break
+                    else:
+                        cur_inst = name_to_instance[cur_inst.task_reuse_scope]
+                        history.add(cur_inst.name)
+
+                tags.append(mapper[inst.name])
+
+        for i in range(1, num_instances):
+            for j in range(i):
+                if tags[i] == tags[j]:
+                    assert instances[i].Paradigm == \
+                            instances[j].Paradigm, \
+                            "paradigm of reuse tasks should be consistent"
+                    instances[i].task_reuse_scope = instances[j].name
+                    break
+
+        self.instances = instances
+        self.mrs = mrs
+        self.Backbone = Backbone
+        self.bb_conf = bb_conf
+        self.bb_name = bb_name
+
+        self.has_init_train = False
+        self.has_init_pred = False
+
+        if self._for_train:
+            print("initialing for training...")
+            self._init_train()
+            self.has_init_train = True
+            
+    def _init_train(self):
+        
+        instances = self.instances
+        Backbone = self.Backbone
+        bb_conf = self.bb_conf
+        bb_name = self.bb_name
+        dev_count = self.dev_count
+        num_instances = len(instances)
+        mrs = self.mrs
+
+        # set first_target/main task instance
+        main_inst = None
+        for inst in instances:
+            if inst.is_target:
+                main_inst = inst
+                inst.is_first_target = True
+                break
+        main_conf = main_inst.config
+        if not os.path.exists(main_conf['save_path']):
+            os.makedirs(main_conf['save_path'])
+            os.makedirs(os.path.join(main_conf['save_path'], 'ckpt'))
+        
+        # prepare backbone
+        train_backbone = Backbone(bb_conf, phase='train')
+        pred_backbone = Backbone(bb_conf, phase='pred')
+
+        # create reader, task
+        # then check i/o across reader, backbone and task_layer
+        task_attrs = []
+        pred_task_attrs = []
+        for inst in instances:
+            train_reader = inst.Reader(inst.config, phase='train')
+            inst.reader['train'] = train_reader
+            train_parad = inst.Paradigm(inst.config, phase='train', backbone_config=bb_conf)
+            inst.task_layer['train'] = train_parad
+            task_attr_from_reader = _encode_inputs(train_parad.inputs_attrs['reader'], inst.name)
+            task_attrs.append(task_attr_from_reader)
+
+            _check_io(train_backbone.inputs_attr, train_reader.outputs_attr, in_name=bb_name+'_backbone', out_name='reader.train')
+            _check_io(train_parad.inputs_attrs['reader'], train_reader.outputs_attr, in_name='task_paradigm.train.reader', out_name='reader.train')
+            _check_io(train_parad.inputs_attrs['backbone'], train_backbone.outputs_attr, in_name='task_paradigm.train.backbone', out_name=bb_name+'_backbone')
+
+            if inst.is_target:
+                if 'pred_file' not in inst.config:
+                    inst.config['pred_file'] = ''
+                pred_reader = inst.Reader(inst.config, phase='pred')
+                pred_parad = inst.Paradigm(inst.config, phase='pred', backbone_config=bb_conf)
+                inst.task_layer['pred'] = pred_parad
+                task_attr_from_reader = _encode_inputs(pred_parad.inputs_attrs['reader'], inst.name)
+                pred_task_attrs.append(task_attr_from_reader)
+                _check_io(pred_backbone.inputs_attr, pred_reader.outputs_attr, in_name=bb_name+'_backbone', out_name='reader.pred')
+                _check_io(pred_parad.inputs_attrs['reader'], pred_reader.outputs_attr, in_name='task_paradigm.pred.reader', out_name='reader.pred')
+                _check_io(pred_parad.inputs_attrs['backbone'], pred_backbone.outputs_attr, in_name='task_paradigm.pred.backbone', out_name=bb_name+'_backbone')
+
+        # merge reader input attrs from backbone and task_instances
+        joint_input_names, joint_shape_and_dtypes, name_to_position = merge_input_attrs(train_backbone.inputs_attr, task_attrs)
+        pred_joint_input_names, pred_joint_shape_and_dtypes, _ = merge_input_attrs(pred_backbone.inputs_attr, pred_task_attrs, insert_taskid=False, insert_batchsize=False, insert_seqlen=False, insert_batchsize_x_seqlen=False)
+        # shapes: [task_id, shapes_of_backbone, shapes_of_inst1, ..., shapes_of_instN]
+
+        if DEBUG:
+            print('----- for debug -----')
+            print('joint input names:')
+            print(joint_input_names)
+            print('joint input shape and dtypes:')
+            print(joint_shape_and_dtypes)
+
+        # load data
+        for inst in instances:
+            print(inst.name+": preparing data...", end='')
+            inst.reader['train'].load_data()
+            print('ok!')
+
+        # merge dataset iterators and create net input vars
+        iterators = []
+        prefixes = []
+        mrs = []
+        for inst in instances:
+            iterators.append(inst.reader['train'].iterator())
+            prefixes.append(inst.name)
+            mrs.append(inst.mix_ratio)
+
+        joint_iterator_fn = create_joint_iterator_fn(iterators, prefixes, joint_shape_and_dtypes, mrs, name_to_position, dev_count=dev_count, verbose=VERBOSE, return_type='dict')
+        self._joint_iterator_fn = joint_iterator_fn
+
+        input_attrs = [[i, j, k] for i, (j,k) in zip(joint_input_names, joint_shape_and_dtypes)]
+        pred_input_attrs = [[i, j, k] for i, (j,k) in zip(pred_joint_input_names, pred_joint_shape_and_dtypes)]
+        # net_inputs = create_net_inputs(input_attrs, async=True, iterator_fn=joint_iterator_fn, dev_count=dev_count, n_prefetch=3)
+        net_inputs = create_net_inputs(input_attrs, async=False)
+        self._net_inputs = net_inputs
+
+        # build backbone and task layers
+        train_prog = fluid.default_main_program()
+        train_init_prog = fluid.default_startup_program()
+        bb_output_vars = train_backbone.build(net_inputs, scope_name='__paddlepalm_')
+        assert sorted(bb_output_vars.keys()) == sorted(train_backbone.outputs_attr.keys())
+        
+        pred_prog = fluid.Program()
+        pred_init_prog = fluid.Program()
+
+        with fluid.program_guard(main_program = pred_prog, startup_program = pred_init_prog):
+            pred_net_inputs = create_net_inputs(pred_input_attrs)
+            pred_bb_output_vars = pred_backbone.build(pred_net_inputs, scope_name='__paddlepalm_')
+
+        fluid.framework.switch_main_program(train_prog)
+        fluid.framework.switch_startup_program(train_init_prog)
+
+        task_output_vars = {}
+        for inst in instances:
+            task_inputs = {'backbone': bb_output_vars}
+            task_inputs_from_reader = _decode_inputs(net_inputs, inst.name)
+            task_inputs['reader'] = task_inputs_from_reader
+
+            scope = inst.task_reuse_scope + '/'
+            with fluid.unique_name.guard(scope):
+                output_vars = inst.build_task_layer(task_inputs, phase='train', scope=scope)
+                output_vars = {inst.name+'/'+key: val for key, val in output_vars.items()}
+                old = len(task_output_vars) # for debug
+                task_output_vars.update(output_vars)
+                assert len(task_output_vars) - old == len(output_vars) # for debug
+
+            # prepare predict vars for saving inference model
+            if inst.is_target:
+
+                with fluid.program_guard(pred_prog, pred_init_prog):
+                    cur_inputs = _decode_inputs(pred_net_inputs, inst.name)
+                    inst.pred_input = cur_inputs
+                    pred_task_inputs = {'backbone': pred_bb_output_vars, 'reader': cur_inputs}
+                    scope = inst.task_reuse_scope + '/'
+                    with fluid.unique_name.guard(scope):
+                        inst.build_task_layer(pred_task_inputs, phase='pred', scope=scope)
+
+
+        bb_fetches = {k: v.name for k,v in bb_output_vars.items()}
+        task_fetches = {k: v.name for k,v in task_output_vars.items()}
+        fetches = task_fetches
+        fetches['__task_id'] = net_inputs['__task_id'].name
+
+        # compute loss
+        task_id_var = net_inputs['__task_id']
+        task_id_vec = fluid.one_hot(task_id_var, num_instances)
+        losses = fluid.layers.concat([task_output_vars[inst.name+'/loss'] for inst in instances], axis=0)
+        loss = layers.reduce_sum(task_id_vec * losses)
+
+        main_reader = main_inst.reader['train']
+
+        num_examples = main_reader.num_examples
+        for inst in instances:
+            max_train_steps = int(main_conf['num_epochs']* inst.mix_ratio * (num_examples // main_conf['batch_size']  // dev_count))
+            if inst.is_target:
+                print('{}: expected train steps {}.'.format(inst.name, max_train_steps))
+            inst.steps_pur_epoch = inst.reader['train'].num_examples // main_conf['batch_size']  // dev_count
+            inst.expected_train_steps = max_train_steps
+
+        global_max_train_steps = int(main_conf['num_epochs'] * sum(mrs) * (num_examples // main_conf['batch_size']  // dev_count))
+        print('Estimated overall train steps {}.'.format(global_max_train_steps))
+
+        if 'warmup_proportion' in main_conf and main_conf['warmup_proportion'] > 0:
+            warmup_steps = int(global_max_train_steps * main_conf['warmup_proportion'])
+            print('Warmup steps: '+str(warmup_steps))
+        else:
+            warmup_steps = 0
+
+        # build optimizer
+        if 'optimizer' in main_conf:
+            optim_mod = importlib.import_module(OPTIMIZER_DIR + '.' + main_conf['optimizer'])
+            optimize = getattr(optim_mod, OPTIMIZE_METHOD)
+            optimize(loss, main_conf, max_train_steps, warmup_steps, fluid.default_main_program())
+
+            loss.persistable = True
+            if main_conf.get('use_ema', False):
+                assert 'ema_decay' in main_conf, "ema_decay should be set when use_ema is enabled."
+                ema = fluid.optimizer.ExponentialMovingAverage(main_conf['ema_decay'])
+                ema.update()
+
+        # prepare for train
+        self.train_backbone = train_backbone
+        self.train_program = fluid.CompiledProgram(fluid.default_main_program()).with_data_parallel(loss_name=loss.name)
+        self.saver_program = fluid.default_main_program()
+
+        self.main_inst = main_inst
+        self.fetches = fetches
+        self.has_init_train = True
+        self.has_init_pred = True
+
+        self.exe.run(fluid.default_startup_program())
+        print("\nRandomly initialize parameters...\n")
+
+    def _init_pred(self, instance, infer_model_path):
+        inst = instance
+        if 'pred_output_path' not in inst.config:
+            inst.config['pred_output_path'] = os.path.join(inst.config.get('save_path', '.'), inst.name)
+
+        if not os.path.exists(inst.config['pred_output_path']):
+            os.makedirs(inst.config['pred_output_path'])
+
+        pred_backbone = self.Backbone(self.bb_conf, phase='pred')
+        pred_parad = inst.Paradigm(inst.config, phase='pred', backbone_config=self.bb_conf)
+        inst.task_layer['pred'] = pred_parad
+        pred_joint_input_names, pred_joint_shape_and_dtypes, name_to_position = merge_input_attrs(
+            pred_backbone.inputs_attr, inst.task_layer['pred'].inputs_attrs['reader'], 
+            insert_taskid=False, insert_batchsize=False, insert_seqlen=False, insert_batchsize_x_seqlen=False)
+
+        pred_prog = inst.load(infer_model_path)
+        if inst.reader['pred'] is None:
+            pred_reader = inst.Reader(inst.config, phase='pred')
+            inst.reader['pred'] = pred_reader
+        return pred_prog
+
+    def load_pretrain(self, pretrain_path=None):
+        # load pretrain model (or ckpt)
+        if pretrain_path is None:
+            assert 'pretrain_path' in self.main_conf, "pretrain_path NOT set."
+            pretrain_path = self.main_conf['pretrain_path']
+
+        init_pretraining_params(
+            self.exe,
+            pretrain_path,
+            main_program=fluid.default_startup_program())
+
+
+    def train(self):
+
+        if not self.has_init_train:
+            self._init_train()
+            self.has_init_train = True
+
+        instances = self.instances
+        num_instances = self.num_instances
+        main_inst = self.main_inst
+        main_conf = main_inst.config
+
+        backbone = self.train_backbone
+        train_program = self.train_program
+        saver_program = self.saver_program
+        fetches = self.fetches
+
+        finish = []
+        for inst in instances:
+            if inst.is_target:
+                if inst.expected_train_steps > 0:
+                    finish.append(False)
+                else:
+                    finish.append(True)
+                    print(inst.name+': train finished!')
+                    inst.save()
+        
+        def train_finish():
+            for inst in instances:
+                if inst.is_target:
+                    if not inst.train_finish:
+                        return False
+            return True
+
+        def pack_multicard_feed(iterator, net_inputs, dev_count):
+            ret = []
+            mask = []
+            for i in range(dev_count):
+                temp = {}
+                content, flag = next(iterator)
+                for q, var in net_inputs.items():
+                    temp[var.name] = content[q]
+                ret.append(temp)
+                mask.append(1 if flag else 0)
+            return ret, mask
+
+        # do training
+        fetch_names, fetch_list = zip(*fetches.items())
+
+        main_step = 0 # only count for main task
+        global_step = 0 # count for all tasks
+        epoch = 0
+        time_begin = time.time()
+        backbone_buffer = []
+
+        def multi_dev_reader(reader, dev_count):
+            def worker(reader, dev_count, queue):
+                dev_batches = []
+                for index, data in enumerate(reader()):
+                    if len(dev_batches) < dev_count:
+                        dev_batches.append(data)
+                    if len(dev_batches) == dev_count:
+                        queue.put((dev_batches, 0))
+                        dev_batches = []
+                # For the prediction of the remained batches, pad more batches to 
+                # the number of devices and the padded samples would be removed in
+                # prediction outputs. 
+                if len(dev_batches) > 0:
+                    num_pad = dev_count - len(dev_batches)
+                    for i in range(len(dev_batches), dev_count):
+                        dev_batches.append(dev_batches[-1])
+                    queue.put((dev_batches, num_pad))
+                queue.put(None)
+
+            queue = Queue.Queue(dev_count*2)
+            p = Thread(
+                target=worker, args=(reader, dev_count, queue))
+            p.daemon = True
+            p.start()
+            while True:
+                ret = queue.get()
+                if ret is not None:
+                    batches, num_pad = ret
+                    queue.task_done()
+                    for batch in batches:
+                        flag = num_pad == 0
+                        if num_pad > 0:
+                            num_pad -= 1
+                        yield batch, flag
+                else:
+                    break
+            queue.join()
+        
+        joint_iterator = multi_dev_reader(self._joint_iterator_fn, self.dev_count)
+        
+        while not train_finish():
+            feed, mask = pack_multicard_feed(joint_iterator, self._net_inputs, self.dev_count)
+            rt_outputs = self.exe.run(train_program, feed=feed, fetch_list=fetch_list)
+            rt_outputs = {k:v for k,v in zip(fetch_names, rt_outputs)}
+            rt_task_id = np.squeeze(rt_outputs['__task_id']).tolist()
+            rt_task_id = rt_task_id[0] if isinstance(rt_task_id, list) else rt_task_id
+            cur_task = instances[rt_task_id]
+
+            backbone_rt_outputs = {k:v for k,v in rt_outputs.items() if '/' not in k}
+            backbone_buffer.append(backbone.postprocess(backbone_rt_outputs))
+            
+            task_rt_outputs = {k[len(cur_task.name+'/'):]: v for k,v in rt_outputs.items() if k.startswith(cur_task.name+'/')}
+            instances[rt_task_id].task_layer['train'].postprocess(task_rt_outputs)
+
+            global_step += 1
+            cur_task.cur_train_step += 1
+
+            cur_task_global_step = cur_task.cur_train_step + cur_task.cur_train_epoch * cur_task.steps_pur_epoch
+            if cur_task.is_target and cur_task.save_infermodel_every_n_steps > 0 and cur_task_global_step % cur_task.save_infermodel_every_n_steps == 0:
+                cur_task.save(suffix='.step'+str(cur_task_global_step))
+
+            if global_step % main_conf.get('print_every_n_steps', 5) == 0:
+                loss = rt_outputs[cur_task.name+'/loss']
+                loss = np.mean(np.squeeze(loss)).tolist()
+
+                time_end = time.time()
+                time_cost = time_end - time_begin
+
+                print("Global step: {}. Task: {}, step {}/{} (epoch {}), loss: {:.3f}, speed: {:.2f} steps/s".format(
+                       global_step, cur_task.name, cur_task.cur_train_step, cur_task.steps_pur_epoch, cur_task.cur_train_epoch,
+                       loss, main_conf.get('print_every_n_steps', 5) / time_cost))
+                time_begin = time.time()
+
+            if cur_task.train_finish and cur_task.cur_train_step + cur_task.cur_train_epoch * cur_task.steps_pur_epoch == cur_task.expected_train_steps:
+                print(cur_task.name+': train finished!')
+                cur_task.save()
+
+            if 'save_ckpt_every_n_steps' in main_conf and global_step % main_conf['save_ckpt_every_n_steps'] == 0:
+                save_path = os.path.join(main_conf['save_path'], 'ckpt', 
+                                         "step_" + str(global_step))
+                fluid.io.save_persistables(self.exe, save_path, saver_program)
+                print('checkpoint has been saved at '+save_path)
+
+        save_path = os.path.join(main_conf['save_path'], 'ckpt',
+                                 "step_" + str(global_step))
+        fluid.io.save_persistables(self.exe, save_path, saver_program)
+        print('checkpoint has been saved at '+save_path)
+
+        print("ALL tasks train finished, exiting...")
+            
+    def pred(self, task_instance, inference_model_dir=None):
+        if self._for_train:
+            raise Exception('This controller is a trainer. Please build a new controller with for_train=False for predicting.')
+
+        assert isinstance(task_instance, str)
+        if isinstance(inference_model_dir, str):
+            assert os.path.exists(inference_model_dir), inference_model_dir+" not found."
+        # if not self.has_init_pred and inference_model_dir is None:
+        #     raise ValueError('infer_model_path is required for prediction.')
+        if inference_model_dir is None:
+            assert 'save_path' in self.mtl_conf, "one of the `inference_model_dir` and 'save_path' should be set to load inference model."
+            inference_model_dir = os.path.join(self.mtl_conf['save_path'], task_instance, 'infer_model')
+
+        instance = None
+        for inst in self.instances:
+            if inst.name == task_instance:
+                instance = inst
+                break
+
+        if instance is None:
+            raise ValueError(task_instance + ' is not a valid task_instance.')
+
+        pred_prog = self._init_pred(instance, inference_model_dir)
+                
+        inst = instance
+        print(inst.name+": loading data...")
+        inst.reader['pred'].load_data()
+        fetch_names, fetch_vars = inst.pred_fetch_list
+
+        print('predicting...')
+        mapper = {k:v for k,v in inst.pred_input}
+        buf = []
+        for feed in inst.reader['pred'].iterator():
+            feed = _encode_inputs(feed, inst.name, cand_set=mapper)
+            feed = {mapper[k]: v for k,v in feed.items()}
+
+            rt_outputs = self.exe.run(pred_prog, feed, fetch_vars)
+            rt_outputs = {k:v for k,v in zip(fetch_names, rt_outputs)}
+            inst.postprocess(rt_outputs, phase='pred')
+        if inst.task_layer['pred'].epoch_inputs_attrs:
+            reader_outputs = inst.reader['pred'].get_epoch_outputs()
+        else:
+            reader_outputs = None
+        inst.epoch_postprocess({'reader':reader_outputs}, phase='pred')
+
+
+if __name__ == '__main__':
+    assert len(sys.argv) == 2, "Usage: python mtl_controller.py <mtl_conf_path>"
+    conf_path = sys.argv[1]
+    del sys.argv[1]
+    controller = Controller(conf_path)
+    if controller.main_conf['do_train']:
+        controller.train()
+
+
+
+__all__ = ["Controller"]
+
+            
+