support different negative sampling strategies

[shenweichen]

1. add deepmatch.utils.NegativeSampler
2. remove deepmatch.layers.core.NegativeSampler
3. add temperature,sampler_config,loss_type for models
4. add some google colab scripts

[codecov]

Codecov Report

Merging #81 (466ae25) into master (be6c028) will decrease coverage by 0.22%.
The diff coverage is 92.12%.

@@            Coverage Diff             @@
##           master      #81      +/-   ##
==========================================
- Coverage   89.13%   88.91%   -0.23%     
==========================================
  Files          14       14              
  Lines         847      911      +64     
==========================================
+ Hits          755      810      +55     
- Misses         92      101       +9     

Flag	Coverage Δ
pytest	88.91% <92.12%> (-0.23%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
deepmatch/layers/sequence.py	76.81% <ø> (ø)
deepmatch/layers/interaction.py	81.89% <33.33%> (ø)
deepmatch/utils.py	80.76% <85.71%> (+1.28%)	⬆️
deepmatch/models/dssm.py	97.05% <92.30%> (-2.95%)	⬇️
deepmatch/models/fm.py	97.05% <92.30%> (-2.95%)	⬇️
deepmatch/layers/core.py	90.66% <93.84%> (-1.13%)	⬇️
deepmatch/__init__.py	100.00% <100.00%> (ø)
deepmatch/layers/__init__.py	100.00% <100.00%> (ø)
deepmatch/models/mind.py	95.18% <100.00%> (+0.05%)	⬆️
deepmatch/models/ncf.py	100.00% <100.00%> (ø)
... and 3 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update be6c028...466ae25. Read the comment docs.