gaphex / bert_experimental Goto Github PK
View Code? Open in Web Editor NEWcode and supplementary materials for a series of Medium articles about the BERT model
License: MIT License
code and supplementary materials for a series of Medium articles about the BERT model
License: MIT License
Hello, I appreciate if you can help in the following issue,
I am using the same logic (same code) to recommend movies, everything seems to be working except when am wanting the result, i got this runtime error " Graph is finalized and cannot be modified issue "
I think that the generator code part in buid_input_fn isn't working fine
I hope if you have an explanation and a solution , i ll be really thankful
Hi i try to use BertLayer encoder and run my model in google colab.
I use tensorflox 1.x
First i clone your github code
!git clone https://github.com/gaphex/bert_experimental
sys.path.insert(0, 'bert_experimental')
from bert_experimental.finetuning.text_preprocessing import build_preprocessor
from bert_experimental.finetuning.bert_layer import BertLayer
from bert_experimental.finetuning.modeling import BertConfig, BertModel, build_bert_module
When i use the encoder in my model
inp1 = tf.keras.Input(shape=(1,), dtype=tf.string)
encoder = BertLayer(bert_path="./bert-module/", seq_len=48,n_tune_layers=3, do_preprocessing=True,
verbose=False, pooling="mean", trainable=True, tune_embeddings=False)
h1 = tf.keras.layers.Dense(50, activation='relu')(encoder(inp1))
I have an error :
NameError Traceback (most recent call last)
in ()
9
10
---> 11 h1 = tf.keras.layers.Dense(50, activation='relu')(encoder(inp1))
12 h1 = tf.keras.layers.Dropout(0.1)(h1)
13
1 frames
/tensorflow-1.15.2/python3.6/tensorflow_core/python/autograph/impl/api.py in wrapper(*args, **kwargs)
235 except Exception as e: # pylint:disable=broad-except
236 if hasattr(e, 'ag_error_metadata'):
--> 237 raise e.ag_error_metadata.to_exception(e)
238 else:
239 raise
NameError: in converted code:
relative to bert_experimental/bert_experimental/finetuning:
bert_layer.py:124 call
pooled = masked_reduce_mean(seq_output, input_mask)
bert_layer.py:314 masked_reduce_mean
return tf.reduce_sum(mul_mask(x, m), axis=1) / (
NameError: name 'mul_mask' is not defined
this code run without error in july, now there is a scope pb with mul_mask
Thanks for your help
Luc
Hi all,
first of all, thank you for a great resource (especially when I get it to work). I am trying to use this approach for fine tunning (NER) in biomedical NLP with BioBERT. Could you maybe point me in the right direction as to how to solve the issues and successfully load BioBERT?
Traceback (most recent call last):
File "/Users/juricaseva/Documents/projects/know-nlp/src/models/ner/Train_BiLSTM.py", line 115, in
model.train_model(corpus_features, splits, evaluation)
File "/Users/juricaseva/Documents/projects/know-nlp/src/models/ner/Train_BiLSTM.py", line 76, in train_model
model = bilstm_model.get_model()
File "/Users/juricaseva/Documents/projects/know-nlp/src/util/neural_models.py", line 150, in get_model
))((encoder(word_input)))
File "/anaconda3/envs/know-nlp-playground/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py", line 824, in call
self._maybe_build(inputs)
File "/anaconda3/envs/know-nlp-playground/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py", line 2146, in _maybe_build
self.build(input_shapes)
File "/Users/juricaseva/Documents/projects/know-nlp/src/util/bert_experimental/finetuning/bert_layer.py", line 34, in build
trainable=self.trainable, name=f"{self.name}_module")
File "/anaconda3/envs/know-nlp-playground/lib/python3.6/site-packages/tensorflow_hub/module.py", line 154, in init
self._spec = as_module_spec(spec)
File "/anaconda3/envs/know-nlp-playground/lib/python3.6/site-packages/tensorflow_hub/module.py", line 36, in as_module_spec
return load_module_spec(spec)
File "/anaconda3/envs/know-nlp-playground/lib/python3.6/site-packages/tensorflow_hub/module.py", line 64, in load_module_spec
return registry.loader(path)
File "/anaconda3/envs/know-nlp-playground/lib/python3.6/site-packages/tensorflow_hub/registry.py", line 45, in call
self._name, args, kwargs))
RuntimeError: Missing implementation that supports: loader(*('/Users/juricaseva/Documents/projects/know-nlp/pretrained/research/biobert_v1.1_pubmed',), **{})
Hi,
Running the colab lets me encounter an issue in step 2: module 'tensorflow' has no attribute 'InteractiveSession'
Any ideas to solve that one to be able to move on?
Hi,
I tried to adjust the code in order to work with TF2. However I'm new in TF and I got stuck and I could not find the alternative for tf.contrib.layers.layer_norm
in TF2, some people said that tf.keras.layers.LayerNormalization
will be equivalent but input parameters are different.
Thanks for your hardworking,
I'm looking forward to seeing the code is compatible on TF2
Hi, first of all, thanks a lot for the effort in the Medium articles on BERT, i found them super helpful for my research
I'm trying to run this notebook, but it does not seem to be working anymore on colab, due to changes on the standardized TF and Python versions (It defaults to 2.1 and 3.7 nowadays). Would you mind giving me the original requirements which were used so i can reproduce the experiment?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.