Upload folder using huggingface_hub

.ipynb_checkpoints/README-checkpoint.md

@@ -0,0 +1,108 @@
+---
+language:
+- en
+- da
+tags:
+- translation
+license: cc-by-4.0
+datasets:
+- quickmt/quickmt-train.da-en
+model-index:
+- name: quickmt-da-en
+  results:
+  - task:
+      name: Translation dan-eng
+      type: translation
+      args: dan-eng
+    dataset:
+      name: flores101-devtest
+      type: flores_101
+      args: dan_Latn eng_Latn devtest
+    metrics:
+    - name: BLEU
+      type: bleu
+      value: 49.02
+    - name: CHRF
+      type: chrf
+      value: 71.78
+    - name: COMET
+      type: comet
+      value: 90.00
+---
+
+
+# `quickmt-da-en` Neural Machine Translation Model 
+
+`quickmt-da-en` is a reasonably fast and reasonably accurate neural machine translation model for translation from `da` into `en`.
+
+
+## Try it on our Huggingface Space
+
+Give it a try before downloading here: https://huggingface.co/spaces/quickmt/QuickMT-Demo
+
+
+## Model Information
+
+* Trained using [`eole`](https://github.com/eole-nlp/eole) 
+* 200M parameter transformer 'big' with 8 encoder layers and 2 decoder layers
+* 32k separate Sentencepiece vocabs
+* Exported for fast inference to [CTranslate2](https://github.com/OpenNMT/CTranslate2) format
+* Training data: https://huggingface.co/datasets/quickmt/quickmt-train.da-en/tree/main
+
+See the `eole` model configuration in this repository for further details and the `eole-model` for the raw `eole` (pytorch) model.
+
+
+## Usage with `quickmt`
+
+You must install the Nvidia cuda toolkit first, if you want to do GPU inference.
+
+Next, install the `quickmt` python library and download the model:
+
+```bash
+git clone https://github.com/quickmt/quickmt.git
+pip install ./quickmt/
+
+quickmt-model-download quickmt/quickmt-da-en ./quickmt-da-en
+```
+
+Finally use the model in python:
+
+```python
+from quickmt import Translator
+
+# Auto-detects GPU, set to "cpu" to force CPU inference
+t = Translator("./quickmt-da-en/", device="auto")
+
+# Translate - set beam size to 1 for faster speed (but lower quality)
+sample_text = 'Dr. Ehud Ur, professor i medicin på Dalhousie University i Halifax, Nova Scotia, og formand for den kliniske og videnskabelige afdeling af Canadian Diabetes Association, advarede om at forskningen stadig er i dens tidlige stadier.'
+
+t(sample_text, beam_size=5)
+```
+
+> 'Dr. Ehud Ur, a professor of medicine at Dalhousie University in Halifax, Nova Scotia, and chairman of the clinical and scientific department of the Canadian Diabetes Association, warned that the research is still in its early stages.'
+
+```python
+# Get alternative translations by sampling
+# You can pass any cTranslate2 `translate_batch` arguments
+t([sample_text], sampling_temperature=1.2, beam_size=1, sampling_topk=50, sampling_topp=0.9)
+```
+
+> 'Dr Ehud Ur, professor of medicine at Dalhousie University in Halifax, Nova Scotia, and chairman of the clinical and scientific branch of the Canadian Diabetes Association, warned that the research is still in its early stages.'
+
+The model is in `ctranslate2` format, and the tokenizers are `sentencepiece`, so you can use `ctranslate2` directly instead of through `quickmt`. It is also possible  to get this model to work with e.g. [LibreTranslate](https://libretranslate.com/) which also uses `ctranslate2` and `sentencepiece`. A model in safetensors format to be used with `eole` is also provided.
+
+
+## Metrics
+
+`bleu` and `chrf2` are calculated with [sacrebleu](https://github.com/mjpost/sacrebleu) on the [Flores200 `devtest` test set](https://huggingface.co/datasets/facebook/flores) ("dan_Latn"->"eng_Latn"). `comet22` with the [`comet`](https://github.com/Unbabel/COMET) library and the [default model](https://huggingface.co/Unbabel/wmt22-comet-da). "Time (s)" is the time in seconds to translate the flores-devtest dataset (1012 sentences) on an RTX 4070s GPU with batch size 32.
+
+## da -> en flores-devtest metrics
+
+|                                  |   bleu |   chrf2 |   comet22 |   Time (s) |
+|:---------------------------------|-------:|--------:|----------:|-----------:|
+| quickmt/quickmt-da-en            |  49.02 |   71.78 |     90    |       1.18 |
+| facebook/nllb-200-distilled-600M |  47.44 |   70.14 |     89.46 |      20.96 |
+| facebook/nllb-200-distilled-1.3B |  49.64 |   71.61 |     90.2  |      36.46 |
+| facebook/m2m100_418M             |  39.23 |   65.42 |     85.85 |      17.51 |
+| facebook/m2m100_1.2B             |  45.24 |   69.44 |     89    |      33.9  |
+

README.md

@@ -1,3 +1,108 @@
----
-license: cc-by-4.0
----
+---
+language:
+- en
+- da
+tags:
+- translation
+license: cc-by-4.0
+datasets:
+- quickmt/quickmt-train.da-en
+model-index:
+- name: quickmt-da-en
+  results:
+  - task:
+      name: Translation dan-eng
+      type: translation
+      args: dan-eng
+    dataset:
+      name: flores101-devtest
+      type: flores_101
+      args: dan_Latn eng_Latn devtest
+    metrics:
+    - name: BLEU
+      type: bleu
+      value: 49.02
+    - name: CHRF
+      type: chrf
+      value: 71.78
+    - name: COMET
+      type: comet
+      value: 90.00
+---
+
+
+# `quickmt-da-en` Neural Machine Translation Model 
+
+`quickmt-da-en` is a reasonably fast and reasonably accurate neural machine translation model for translation from `da` into `en`.
+
+
+## Try it on our Huggingface Space
+
+Give it a try before downloading here: https://huggingface.co/spaces/quickmt/QuickMT-Demo
+
+
+## Model Information
+
+* Trained using [`eole`](https://github.com/eole-nlp/eole) 
+* 200M parameter transformer 'big' with 8 encoder layers and 2 decoder layers
+* 32k separate Sentencepiece vocabs
+* Exported for fast inference to [CTranslate2](https://github.com/OpenNMT/CTranslate2) format
+* Training data: https://huggingface.co/datasets/quickmt/quickmt-train.da-en/tree/main
+
+See the `eole` model configuration in this repository for further details and the `eole-model` for the raw `eole` (pytorch) model.
+
+
+## Usage with `quickmt`
+
+You must install the Nvidia cuda toolkit first, if you want to do GPU inference.
+
+Next, install the `quickmt` python library and download the model:
+
+```bash
+git clone https://github.com/quickmt/quickmt.git
+pip install ./quickmt/
+
+quickmt-model-download quickmt/quickmt-da-en ./quickmt-da-en
+```
+
+Finally use the model in python:
+
+```python
+from quickmt import Translator
+
+# Auto-detects GPU, set to "cpu" to force CPU inference
+t = Translator("./quickmt-da-en/", device="auto")
+
+# Translate - set beam size to 1 for faster speed (but lower quality)
+sample_text = 'Dr. Ehud Ur, professor i medicin på Dalhousie University i Halifax, Nova Scotia, og formand for den kliniske og videnskabelige afdeling af Canadian Diabetes Association, advarede om at forskningen stadig er i dens tidlige stadier.'
+
+t(sample_text, beam_size=5)
+```
+
+> 'Dr. Ehud Ur, a professor of medicine at Dalhousie University in Halifax, Nova Scotia, and chairman of the clinical and scientific department of the Canadian Diabetes Association, warned that the research is still in its early stages.'
+
+```python
+# Get alternative translations by sampling
+# You can pass any cTranslate2 `translate_batch` arguments
+t([sample_text], sampling_temperature=1.2, beam_size=1, sampling_topk=50, sampling_topp=0.9)
+```
+
+> 'Dr Ehud Ur, professor of medicine at Dalhousie University in Halifax, Nova Scotia, and chairman of the clinical and scientific branch of the Canadian Diabetes Association, warned that the research is still in its early stages.'
+
+The model is in `ctranslate2` format, and the tokenizers are `sentencepiece`, so you can use `ctranslate2` directly instead of through `quickmt`. It is also possible  to get this model to work with e.g. [LibreTranslate](https://libretranslate.com/) which also uses `ctranslate2` and `sentencepiece`. A model in safetensors format to be used with `eole` is also provided.
+
+
+## Metrics
+
+`bleu` and `chrf2` are calculated with [sacrebleu](https://github.com/mjpost/sacrebleu) on the [Flores200 `devtest` test set](https://huggingface.co/datasets/facebook/flores) ("dan_Latn"->"eng_Latn"). `comet22` with the [`comet`](https://github.com/Unbabel/COMET) library and the [default model](https://huggingface.co/Unbabel/wmt22-comet-da). "Time (s)" is the time in seconds to translate the flores-devtest dataset (1012 sentences) on an RTX 4070s GPU with batch size 32.
+
+## da -> en flores-devtest metrics
+
+|                                  |   bleu |   chrf2 |   comet22 |   Time (s) |
+|:---------------------------------|-------:|--------:|----------:|-----------:|
+| quickmt/quickmt-da-en            |  49.02 |   71.78 |     90    |       1.18 |
+| facebook/nllb-200-distilled-600M |  47.44 |   70.14 |     89.46 |      20.96 |
+| facebook/nllb-200-distilled-1.3B |  49.64 |   71.61 |     90.2  |      36.46 |
+| facebook/m2m100_418M             |  39.23 |   65.42 |     85.85 |      17.51 |
+| facebook/m2m100_1.2B             |  45.24 |   69.44 |     89    |      33.9  |
+

config.json

@@ -0,0 +1,10 @@
+{
+  "add_source_bos": false,
+  "add_source_eos": false,
+  "bos_token": "<s>",
+  "decoder_start_token": "<s>",
+  "eos_token": "</s>",
+  "layer_norm_epsilon": 1e-06,
+  "multi_query_attention": false,
+  "unk_token": "<unk>"
+}

eole-config.yaml

@@ -0,0 +1,98 @@
+## IO
+save_data: data
+overwrite: True
+seed: 1234
+report_every: 100
+valid_metrics: ["BLEU"]
+tensorboard: true
+tensorboard_log_dir: tensorboard
+
+### Vocab
+src_vocab: da.eole.vocab
+tgt_vocab: en.eole.vocab
+src_vocab_size: 32000
+tgt_vocab_size: 32000
+vocab_size_multiple: 8
+share_vocab: false
+n_sample: 0
+
+data:
+    corpus_1:
+        # path_src: hf://quickmt/quickmt-train.da-en/da
+        # path_tgt: hf://quickmt/quickmt-train.da-en/en
+        # path_sco: hf://quickmt/quickmt-train.da-en/sco
+        path_src: train.da
+        path_tgt: train.en
+    valid:
+        path_src: valid.da
+        path_tgt: valid.en
+
+transforms: [sentencepiece, filtertoolong]
+transforms_configs:
+  sentencepiece:
+    src_subword_model: "da.spm.model"
+    tgt_subword_model: "en.spm.model"
+  filtertoolong:
+    src_seq_length: 256
+    tgt_seq_length: 256
+
+training:
+    # Run configuration
+    model_path: quickmt-da-en-eole-model
+    #train_from: model
+    keep_checkpoint: 4
+    train_steps: 100000
+    save_checkpoint_steps: 5000
+    valid_steps: 5000
+    
+    # Train on a single GPU
+    world_size: 1
+    gpu_ranks: [0]
+
+    # Batching 10240
+    batch_type: "tokens"
+    batch_size: 6000
+    valid_batch_size: 2048
+    batch_size_multiple: 8
+    accum_count: [20]
+    accum_steps: [0]
+
+    # Optimizer & Compute
+    compute_dtype: "fp16"
+    optim: "adamw"
+    #use_amp: False
+    learning_rate: 2.0
+    warmup_steps: 2000
+    decay_method: "noam"
+    adam_beta2: 0.998
+
+    # Data loading
+    bucket_size: 128000
+    num_workers: 4
+    prefetch_factor: 32
+    
+    # Hyperparams
+    dropout_steps: [0]
+    dropout: [0.1]
+    attention_dropout: [0.1]
+    max_grad_norm: 0
+    label_smoothing: 0.1
+    average_decay: 0.0001
+    param_init_method: xavier_uniform
+    normalization: "tokens"
+
+model:
+    architecture: "transformer"
+    share_embeddings: false
+    share_decoder_embeddings: true
+    hidden_size: 1024
+    encoder:
+        layers: 8
+    decoder:
+        layers: 2
+    heads: 8
+    transformer_ff: 4096
+    embeddings:
+        word_vec_size: 1024
+        position_encoding_type: "SinusoidalInterleaved"
+

eole-model/config.json

@@ -0,0 +1,132 @@
+{
+  "tgt_vocab_size": 32000,
+  "share_vocab": false,
+  "tensorboard": true,
+  "n_sample": 0,
+  "src_vocab_size": 32000,
+  "src_vocab": "da.eole.vocab",
+  "transforms": [
+    "sentencepiece",
+    "filtertoolong"
+  ],
+  "report_every": 100,
+  "valid_metrics": [
+    "BLEU"
+  ],
+  "tensorboard_log_dir_dated": "tensorboard/Oct-28_23-01-02",
+  "save_data": "data",
+  "overwrite": true,
+  "tensorboard_log_dir": "tensorboard",
+  "tgt_vocab": "en.eole.vocab",
+  "seed": 1234,
+  "vocab_size_multiple": 8,
+  "training": {
+    "param_init_method": "xavier_uniform",
+    "train_steps": 100000,
+    "attention_dropout": [
+      0.1
+    ],
+    "normalization": "tokens",
+    "model_path": "quickmt-da-en-eole-model",
+    "average_decay": 0.0001,
+    "num_workers": 0,
+    "dropout_steps": [
+      0
+    ],
+    "prefetch_factor": 32,
+    "learning_rate": 2.0,
+    "compute_dtype": "torch.float16",
+    "valid_steps": 5000,
+    "warmup_steps": 2000,
+    "adam_beta2": 0.998,
+    "accum_count": [
+      20
+    ],
+    "batch_type": "tokens",
+    "batch_size_multiple": 8,
+    "bucket_size": 128000,
+    "save_checkpoint_steps": 5000,
+    "keep_checkpoint": 4,
+    "batch_size": 6000,
+    "dropout": [
+      0.1
+    ],
+    "optim": "adamw",
+    "accum_steps": [
+      0
+    ],
+    "valid_batch_size": 2048,
+    "max_grad_norm": 0.0,
+    "gpu_ranks": [
+      0
+    ],
+    "label_smoothing": 0.1,
+    "decay_method": "noam",
+    "world_size": 1
+  },
+  "model": {
+    "position_encoding_type": "SinusoidalInterleaved",
+    "architecture": "transformer",
+    "share_embeddings": false,
+    "heads": 8,
+    "hidden_size": 1024,
+    "transformer_ff": 4096,
+    "share_decoder_embeddings": true,
+    "embeddings": {
+      "position_encoding_type": "SinusoidalInterleaved",
+      "word_vec_size": 1024,
+      "tgt_word_vec_size": 1024,
+      "src_word_vec_size": 1024
+    },
+    "decoder": {
+      "position_encoding_type": "SinusoidalInterleaved",
+      "tgt_word_vec_size": 1024,
+      "heads": 8,
+      "layers": 2,
+      "n_positions": null,
+      "hidden_size": 1024,
+      "decoder_type": "transformer",
+      "transformer_ff": 4096
+    },
+    "encoder": {
+      "position_encoding_type": "SinusoidalInterleaved",
+      "src_word_vec_size": 1024,
+      "heads": 8,
+      "encoder_type": "transformer",
+      "layers": 8,
+      "n_positions": null,
+      "hidden_size": 1024,
+      "transformer_ff": 4096
+    }
+  },
+  "transforms_configs": {
+    "sentencepiece": {
+      "tgt_subword_model": "${MODEL_PATH}/en.spm.model",
+      "src_subword_model": "${MODEL_PATH}/da.spm.model"
+    },
+    "filtertoolong": {
+      "tgt_seq_length": 256,
+      "src_seq_length": 256
+    }
+  },
+  "data": {
+    "corpus_1": {
+      "path_align": null,
+      "path_tgt": "train.en",
+      "path_src": "train.da",
+      "transforms": [
+        "sentencepiece",
+        "filtertoolong"
+      ]
+    },
+    "valid": {
+      "path_align": null,
+      "path_tgt": "valid.en",
+      "path_src": "valid.da",
+      "transforms": [
+        "sentencepiece",
+        "filtertoolong"
+      ]
+    }
+  }
+}

eole-model/da.spm.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:688ecf3cf51d2cf7497720ba36e5f3c6bdfea75630c2c2c8afcf8a0de3580d06
+size 814697

eole-model/en.spm.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8eda8069d07624408f632a155bbfb6db9794b8bb7bdbc9040c8ec16d499b6ef2
+size 800426

eole-model/model.00.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6acf5dd3d4a3f8efe9f32ffa635944452f9c3361fe63de85e33453d4e076af03
+size 840314816

model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8e3d7ae48cd5fc340ae381477206a6fcefc1f2329364d1b22d108373b903e6d
+size 409915789

src.spm.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:688ecf3cf51d2cf7497720ba36e5f3c6bdfea75630c2c2c8afcf8a0de3580d06
+size 814697

tgt.spm.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8eda8069d07624408f632a155bbfb6db9794b8bb7bdbc9040c8ec16d499b6ef2
+size 800426