Comments (4)
Here are some specs about the DPR biencoder training
batch_size: 128
dev_batch_size: 64
adam_eps: 1e-8
adam_betas: (0.9, 0.999)
max_grad_norm: 2.0
log_batch_step: 1
train_rolling_loss_step: 20
weight_decay: 0.0
learning_rate: 2e-5
# Linear warmup over warmup_steps.
warmup_steps: 200
# Number of updates steps to accumulate before performing a backward/update pass.
gradient_accumulation_steps: 1
# Total number of training epochs to perform.
num_train_epochs: 50
eval_per_epoch: 1
hard_negatives: 1
other_negatives: 0
val_av_rank_hard_neg: 30
val_av_rank_other_neg: 30
val_av_rank_bsz: 128
val_av_rank_max_qs: 10000
from multidoc2dial.
Hi Yawei, thank you for the question!
As indicated in README, we used Facebook DPR, specifically
git checkout 49e5838f94ffced8392be750ded2a8fa4a14b5cf
with default configurations including the effective batch size of 128. Please refer to their official scripts for more details.
We provided the script for creating positive and negative examples for MultiDoc2Dial.
Please let us know if you have any questions. Thanks!
from multidoc2dial.
Thanks for sharing the parameters. Now I can reproduce the results in Table 5.
from multidoc2dial.
Is it possible to reproduce the results with sivasankalpp/dpr-multidoc2dial-structure-question
and sivasankalpp/dpr-multidoc2dial-structure-ctx-encoder
checkpoints?
Using the following leads to significantly lower result:
tokenizer_contex = AutoTokenizer.from_pretrained("sivasankalpp/dpr-multidoc2dial-structure-ctx-encoder")
tokenizer_question = AutoTokenizer.from_pretrained("sivasankalpp/dpr-multidoc2dial-structure-ctx-encoder")
model_context = DPRContextEncoder.from_pretrained("sivasankalpp/dpr-multidoc2dial-structure-ctx-encoder")
model_question = AutoModel.from_pretrained("sivasankalpp/dpr-multidoc2dial-structure-ctx-encoder")
input_ids = tokenizer_question(inp_q, return_tensors='pt', truncation=True)["input_ids"]
query_emb = model_question(input_ids).pooler_output
input_ids = tokenizer_contex(inp_p, truncation=True, return_tensors='pt', )["input_ids"]
passage_emb = model_context(input_ids).pooler_output
torch.dot(query_emb, passage_emb)
from multidoc2dial.
Related Issues (12)
- Error in running converter HOT 1
- Performance on BM25 retrieval baseline HOT 4
- How to generate prediction file for sharetask? HOT 9
- Question about data preprocessing HOT 1
- the link for script run_sharedtask_eval.sh is corrupted HOT 1
- no run_finetune_rag.sh and missing positional argument: 'logits_processor' HOT 2
- Question about using multiple gpus
- About data download HOT 4
- Sharing unseen-domain data HOT 2
- question about data preprocessing HOT 5
- -nq model results and n_docs HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from multidoc2dial.