Max Tian

I think it would be useful to use a callback to save the best model in training. The usage of callbacks is described in https://stable-baselines.readthedocs.io/en/master/guide/examples.html#using-callback-monitoring-training ``` best_mean_reward, n_steps = -np.inf,...

How to use BatchLossFilter callback with blurr

2

I've been trying to experiment with using tsai's [BatchLossFilter callback](https://timeseriesai.github.io/tsai/callback.core#BatchLossFilter). If I try to run the training with this callback ``` model = HF_BaseModelWrapper(hf_model) learn = Learner(dls, model, loss_func=LabelSmoothingCrossEntropyFlat(), metrics=[accuracy],...

### System Info tested on multiple versions - `transformers` version: 4.12.3 - Platform: Linux-4.14.281-212.502.amzn2.x86_64-x86_64-with-glibc2.10 - Python version: 3.8.10 - PyTorch version (GPU?): 1.11.0+cu102 (True) - Tensorflow version (GPU?): not installed...

Error when doing shap dependence plot of a single variable

5

The model is a RandomForestClassifier from sklearn, when I try to create the dependence plot from a single value: `shap.dependence_plot("S", shap_values, x_train) ` I get the following error: ``` ---------------------------------------------------------------------------...

I'm trying to implement the res2net in a different way, following how fastai wrote their xresnet. The res2block I have is ``` def no_op(x): return x # no operations done...

MonoT5 .predict gives NaN

2

Trying to use MonoT5 3B on some custom reranking tasks, the gist of the code is ``` model_name = "castorini/monot5-3b-msmarco-10k" cross_encoder_model = MonoT5(model_name, token_false='▁false', token_true='▁true') query, doc = "some query",...

TypeError: can't multiply sequence by non-int of type 'float' in _wide_layer

1

I get this issue with using the _wide_layer ``` 30 def make_layer(self, block, out_channels, n_blocks, dropout_rate, stride): ---> 31 strides = [stride] + [1]*(n_blocks-1) 32 layers = [] 33 TypeError:...