Upload folder using huggingface_hub

README.md

@@ -0,0 +1,277 @@
+---
+language: rna
+tags:
+  - Biology
+  - RNA
+license: agpl-3.0
+datasets:
+  - multimolecule/rnacentral
+library_name: multimolecule
+pipeline_tag: fill-mask
+mask_token: "<mask>"
+widget:
+  - example_title: "HIV-1"
+    text: "GGUC<mask>CUCUGGUUAGACCAGAUCUGAGCCU"
+    output:
+      - label: "<eos>"
+        score: 0.19942431151866913
+      - label: "I"
+        score: 0.1465310901403427
+      - label: "*"
+        score: 0.1448192000389099
+      - label: "<unk>"
+        score: 0.14174020290374756
+      - label: "<cls>"
+        score: 0.13194777071475983
+  - example_title: "microRNA-21"
+    text: "UAGC<mask>UAUCAGACUGAUGUUG"
+    output:
+      - label: "<eos>"
+        score: 0.19946657121181488
+      - label: "I"
+        score: 0.14641942083835602
+      - label: "*"
+        score: 0.14452320337295532
+      - label: "<unk>"
+        score: 0.14180712401866913
+      - label: "<cls>"
+        score: 0.13223469257354736
+---
+
+# ncRNABert
+
+Pre-trained model on non-coding RNA (ncRNA) using a masked language modeling (MLM) objective.
+
+## Disclaimer
+
+This is an UNOFFICIAL implementation of the [A 5’ UTR Language Model for Decoding Untranslated Regions of mRNA and Function Predictions](https://doi.org/10.1101/2023.10.11.561938) by Yanyi Chu, Dan Yu, et al.
+
+The OFFICIAL repository of ncRNABert is at [wangleiofficial/ncRNABert](https://github.com/wangleiofficial/ncRNABert).
+
+> [!WARNING]
+> The MultiMolecule team is aware of a potential risk in reproducing the results of ncRNABert.
+>
+> The ncRNABert apply `softmax` in the `-2` dimension when computing the attention probs. This makes the output of `attention_probs @ value_layer` unreliable when the input sequences are not of the same length (i.e., have padding tokens).
+> MultiMolecule applied a workaround to ensure that the attention masks are applied correctly, but this may lead to different results compared to the original implementation.
+
+> [!CAUTION]
+> The MultiMolecule team is aware of a potential risk in reproducing the results of RibonanzaNet.
+>
+> The original implementation of ncRNABert does not prepend `<bos>` (`<cls>`) and append `<eos>` tokens to the input sequence.
+> This should not affect the performance of the model in most cases, but it can lead to unexpected behavior in some cases.
+>
+> Please set `bos_token=None, eos_token=None` in the tokenizer and set `bos_token_id=None, eos_token_id=None` in the model configuration if you want the exact behavior of the original implementation.
+
+> [!TIP]
+> The MultiMolecule team has confirmed that the provided model and checkpoints are producing the same intermediate representations as the original implementation.
+
+**The team releasing ncRNABert did not write this model card for this model so this model card has been written by the MultiMolecule team.**
+
+## Model Details
+
+ncRNABert is a [bert](https://huggingface.co/google-bert/bert-base-uncased)-style model pre-trained on a large corpus of 5’ untranslated regions (5’UTRs) in a self-supervised fashion. This means that the model was trained on the raw nucleotides of RNA sequences only, with an automatic process to generate inputs and labels from those texts. Please refer to the [Training Details](#training-details) section for more information on the training process.
+
+### Variants
+
+- **[multimolecule/ncrnabert](https://huggingface.co/multimolecule/ncrnabert)**: The ncRNABert model pre-trained on single nucleotide data.
+- **[multimolecule/ncrnabert-3mer](https://huggingface.co/multimolecule/ncrnabert-3mer)**: The ncRNABert model pre-trained on 3-mer data.
+
+### Model Specification
+
+<table>
+<thead>
+  <tr>
+    <th>Variants</th>
+    <th>Num Layers</th>
+    <th>Hidden Size</th>
+    <th>Num Heads</th>
+    <th>Intermediate Size</th>
+    <th>Num Parameters (M)</th>
+    <th>FLOPs (G)</th>
+    <th>MACs (G)</th>
+    <th>Max Num Tokens</th>
+  </tr>
+</thead>
+<tbody>
+  <tr>
+    <td>ncRNABert</td>
+    <td rowspan="2">24</td>
+    <td rowspan="2">1024</td>
+    <td rowspan="2">16</td>
+    <td rowspan="2">4096</td>
+    <td rowspan="2">303.31</td>
+    <td rowspan="2">78.96</td>
+    <td rowspan="2">39.46</td>
+    <td rowspan="2">512</td>
+  </tr>
+  <tr>
+    <td>ncRNABert-3mer</td>
+  </tr>
+</tbody>
+</table>
+
+### Links
+
+- **Code**: [multimolecule.ncrnabert](https://github.com/DLS5-Omics/multimolecule/tree/master/multimolecule/models/ncrnabert)
+- **Data**: [multimolecule/rnacentral](https://huggingface.co/datasets/multimolecule/rnacentral)
+- **Model type**: [BERT](https://huggingface.co/google-bert/bert-base-uncased)
+- **Original Repository**: [wangleiofficial/ncRNABert](https://github.com/wangleiofficial/ncRNABert)
+
+## Usage
+
+The model file depends on the [`multimolecule`](https://multimolecule.danling.org) library. You can install it using pip:
+
+```bash
+pip install multimolecule
+```
+
+### Direct Use
+
+#### Masked Language Modeling
+
+You can use this model directly with a pipeline for masked language modeling:
+
+```python
+>>> import multimolecule  # you must import multimolecule to register models
+>>> from transformers import pipeline
+
+>>> unmasker = pipeline("fill-mask", model="multimolecule/ncrnabert")
+>>> unmasker("gguc<mask>cucugguuagaccagaucugagccu")
+[{'score': 0.19942431151866913,
+  'token': 2,
+  'token_str': '<eos>',
+  'sequence': 'G G U C C U C U G G U U A G A C C A G A U C U G A G C C U'},
+ {'score': 0.1465310901403427,
+  'token': 25,
+  'token_str': 'I',
+  'sequence': 'G G U C I C U C U G G U U A G A C C A G A U C U G A G C C U'},
+ {'score': 0.1448192000389099,
+  'token': 23,
+  'token_str': '*',
+  'sequence': 'G G U C * C U C U G G U U A G A C C A G A U C U G A G C C U'},
+ {'score': 0.14174020290374756,
+  'token': 3,
+  'token_str': '<unk>',
+  'sequence': 'G G U C C U C U G G U U A G A C C A G A U C U G A G C C U'},
+ {'score': 0.13194777071475983,
+  'token': 1,
+  'token_str': '<cls>',
+  'sequence': 'G G U C C U C U G G U U A G A C C A G A U C U G A G C C U'}]
+```
+
+### Downstream Use
+
+#### Extract Features
+
+Here is how to use this model to get the features of a given sequence in PyTorch:
+
+```python
+from multimolecule import RnaTokenizer, NcRnaBertModel
+
+
+tokenizer = RnaTokenizer.from_pretrained("multimolecule/ncrnabert")
+model = NcRnaBertModel.from_pretrained("multimolecule/ncrnabert")
+
+text = "UAGCUUAUCAGACUGAUGUUG"
+input = tokenizer(text, return_tensors="pt")
+
+output = model(**input)
+```
+
+#### Sequence Classification / Regression
+
+> [!NOTE]
+> This model is not fine-tuned for any specific task. You will need to fine-tune the model on a downstream task to use it for sequence classification or regression.
+
+Here is how to use this model as backbone to fine-tune for a sequence-level task in PyTorch:
+
+```python
+import torch
+from multimolecule import RnaTokenizer, NcRnaBertForSequencePrediction
+
+
+tokenizer = RnaTokenizer.from_pretrained("multimolecule/ncrnabert")
+model = NcRnaBertForSequencePrediction.from_pretrained("multimolecule/ncrnabert")
+
+text = "UAGCUUAUCAGACUGAUGUUG"
+input = tokenizer(text, return_tensors="pt")
+label = torch.tensor([1])
+
+output = model(**input, labels=label)
+```
+
+#### Token Classification / Regression
+
+> [!NOTE]
+> This model is not fine-tuned for any specific task. You will need to fine-tune the model on a downstream task to use it for token classification or regression.
+
+Here is how to use this model as backbone to fine-tune for a nucleotide-level task in PyTorch:
+
+```python
+import torch
+from multimolecule import RnaTokenizer, NcRnaBertForTokenPrediction
+
+
+tokenizer = RnaTokenizer.from_pretrained("multimolecule/ncrnabert")
+model = NcRnaBertForTokenPrediction.from_pretrained("multimolecule/ncrnabert")
+
+text = "UAGCUUAUCAGACUGAUGUUG"
+input = tokenizer(text, return_tensors="pt")
+label = torch.randint(2, (len(text), ))
+
+output = model(**input, labels=label)
+```
+
+#### Contact Classification / Regression
+
+> [!NOTE]
+> This model is not fine-tuned for any specific task. You will need to fine-tune the model on a downstream task to use it for contact classification or regression.
+
+Here is how to use this model as backbone to fine-tune for a contact-level task in PyTorch:
+
+```python
+import torch
+from multimolecule import RnaTokenizer, NcRnaBertForContactPrediction
+
+
+tokenizer = RnaTokenizer.from_pretrained("multimolecule/ncrnabert")
+model = NcRnaBertForContactPrediction.from_pretrained("multimolecule/ncrnabert")
+
+text = "UAGCUUAUCAGACUGAUGUUG"
+input = tokenizer(text, return_tensors="pt")
+label = torch.randint(2, (len(text), len(text)))
+
+output = model(**input, labels=label)
+```
+
+## Training Details
+
+ncRNABert used Masked Language Modeling (MLM) as the pre-training objective: taking a sequence, the model randomly masks 15% of the tokens in the input then runs the entire masked sentence through the model and has to predict the masked tokens. This is comparable to the Cloze task in language modeling.
+
+### Training Data
+
+The ncRNABert model was pre-trained on [RNAcentral](https://multimolecule.danling.org/datasets/rnacentral).
+RNAcentral is a free, public resource that offers integrated access to a comprehensive and up-to-date set of non-coding RNA sequences provided by a collaborating group of [Expert Databases](https://rnacentral.org/expert-databases) representing a broad range of organisms and RNA types.
+
+### Training Procedure
+
+#### Preprocessing
+
+ncRNABert used masked language modeling (MLM) as one of the pre-training objectives. The masking procedure is similar to the one used in BERT:
+
+- 15% of the tokens are masked.
+- In 80% of the cases, the masked tokens are replaced by `<mask>`.
+- In 10% of the cases, the masked tokens are replaced by a random token (different) from the one they replace.
+- In the 10% remaining cases, the masked tokens are left as is.
+
+## Contact
+
+Please use GitHub issues of [MultiMolecule](https://github.com/DLS5-Omics/multimolecule/issues) for any questions or comments on the model card.
+
+## License
+
+This model is licensed under the [AGPL-3.0 License](https://www.gnu.org/licenses/agpl-3.0.html).
+
+```spdx
+SPDX-License-Identifier: AGPL-3.0-or-later
+```

config.json

@@ -0,0 +1,34 @@
+{
+  "architectures": [
+    "NcRnaBertForPreTraining"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "head": null,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
+  "id2label": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "label2id": null,
+  "layer_norm_eps": 1e-12,
+  "lm_head": null,
+  "mask_token_id": 4,
+  "max_position_embeddings": 512,
+  "model_type": "ncrnabert",
+  "nmers": 3,
+  "null_token_id": 5,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "num_labels": 1,
+  "pad_token_id": 0,
+  "position_embedding_type": "rotary",
+  "torch_dtype": "float32",
+  "transformers_version": "4.50.0",
+  "type_vocab_size": 2,
+  "unk_token_id": 3,
+  "use_cache": true,
+  "vocab_size": 131
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17648cf1d927f79de9560315ba36b17955e9801aa77b84ed119f20159184e55d
+size 1213738532

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a734652981de277744254be38a9eafda28bf45e8b03da5fb40d593fd2febbd08
+size 1213810986

special_tokens_map.json

@@ -0,0 +1,12 @@
+{
+  "additional_special_tokens": [
+    "<null>"
+  ],
+  "bos_token": "<cls>",
+  "cls_token": "<cls>",
+  "eos_token": "<eos>",
+  "mask_token": "<mask>",
+  "pad_token": "<pad>",
+  "sep_token": "<eos>",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,69 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<cls>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<null>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<null>"
+  ],
+  "bos_token": "<cls>",
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "<cls>",
+  "codon": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "mask_token": "<mask>",
+  "model_max_length": 512,
+  "nmers": 3,
+  "pad_token": "<pad>",
+  "replace_T_with_U": true,
+  "sep_token": "<eos>",
+  "tokenizer_class": "RnaTokenizer",
+  "unk_token": "<unk>"
+}

vocab.txt

@@ -0,0 +1,131 @@
+<pad>
+<cls>
+<eos>
+<unk>
+<mask>
+<null>
+AAA
+AAC
+AAG
+AAU
+AAN
+ACA
+ACC
+ACG
+ACU
+ACN
+AGA
+AGC
+AGG
+AGU
+AGN
+AUA
+AUC
+AUG
+AUU
+AUN
+ANA
+ANC
+ANG
+ANU
+ANN
+CAA
+CAC
+CAG
+CAU
+CAN
+CCA
+CCC
+CCG
+CCU
+CCN
+CGA
+CGC
+CGG
+CGU
+CGN
+CUA
+CUC
+CUG
+CUU
+CUN
+CNA
+CNC
+CNG
+CNU
+CNN
+GAA
+GAC
+GAG
+GAU
+GAN
+GCA
+GCC
+GCG
+GCU
+GCN
+GGA
+GGC
+GGG
+GGU
+GGN
+GUA
+GUC
+GUG
+GUU
+GUN
+GNA
+GNC
+GNG
+GNU
+GNN
+UAA
+UAC
+UAG
+UAU
+UAN
+UCA
+UCC
+UCG
+UCU
+UCN
+UGA
+UGC
+UGG
+UGU
+UGN
+UUA
+UUC
+UUG
+UUU
+UUN
+UNA
+UNC
+UNG
+UNU
+UNN
+NAA
+NAC
+NAG
+NAU
+NAN
+NCA
+NCC
+NCG
+NCU
+NCN
+NGA
+NGC
+NGG
+NGU
+NGN
+NUA
+NUC
+NUG
+NUU
+NUN
+NNA
+NNC
+NNG
+NNU
+NNN