Upload 2 files

README.md

@@ -1,3 +1,156 @@
 ---
 license: apache-2.0
+base_model: arcee-ai/Trinity-Mini
+library_name: peft
+pipeline_tag: text-generation
+tags:
+  - lora
+  - peft
+  - grpo
+  - reinforcement-learning
+  - biomedical
+  - relation-extraction
+  - drug-protein
+  - moe
+language:
+  - en
 ---
+
+<p align="center">
+  <img src="assets/logo.png" alt="Trinity-Mini-DrugProt-Think" width="350" />
+</p>
+
+<p align="center">
+  <strong>Trinity-Mini-DrugProt-Think</strong><br/>
+  RLVR (GRPO) + LoRA post-training on Arcee Trinity Mini for DrugProt relation classification.
+</p>
+
+<p align="center">
+  <a href="index.html">📝 <strong>Report</strong></a> &nbsp; | &nbsp;
+  <a href="https://medium.com/@jakimovski_bojan/9e1c1c430ce9"><img src="https://www.sysgroup.com/wp-content/uploads/2025/02/Amazon_Web_Services-Logo.wine_.png" height="20" style="vertical-align:middle;"/> <strong>AWS deployment guide</strong></a> &nbsp; | &nbsp;
+  <a href="https://github.com/LokaHQ/Trinity-Mini-DrugProt-Think" aria-label="GitHub"><svg viewBox="0 0 16 16" fill="currentColor" width="20" height="20" style="vertical-align:middle;"><path d="M8 0C3.58 0 0 3.58 0 8c0 3.54 2.29 6.53 5.47 7.59.4.07.55-.17.55-.38 0-.19-.01-.82-.01-1.49-2.01.37-2.53-.49-2.69-.94-.09-.23-.48-.94-.82-1.13-.28-.15-.68-.52-.01-.53.63-.01 1.08.58 1.23.82.72 1.21 1.87.87 2.33.66.07-.52.28-.87.51-1.07-1.78-.2-3.64-.89-3.64-3.95 0-.87.31-1.59.82-2.15-.08-.2-.36-1.02.08-2.12 0 0 .67-.21 2.2.82.64-.18 1.32-.27 2-.27s1.36.09 2 .27c1.53-1.04 2.2-.82 2.2-.82.44 1.1.16 1.92.08 2.12.51.56.82 1.27.82 2.15 0 3.07-1.87 3.75-3.65 3.95.29.25.54.73.54 1.48 0 1.07-.01 1.93-.01 2.2 0 .21.15.46.55.38A8.01 8.01 0 0 0 16 8c0-4.42-3.58-8-8-8z"/></svg> <strong>GitHub</strong></a>
+</p>
+
+
+# Trinity-Mini-DrugProt-Think
+
+A LoRA adapter fine-tuned on [Arcee Trinity Mini](https://huggingface.co/arcee-ai/Trinity-Mini) using GRPO (Group Relative Policy Optimization) for **drug-protein relation extraction** on the [DrugProt (BioCreative VII)](https://huggingface.co/datasets/OpenMed/drugprot-parquet) benchmark. The model classifies 13 types of drug-protein interactions from PubMed abstracts, producing structured pharmacological reasoning traces before giving its answer.
+
+📄 **Blog post:** [Post-Training an Open MoE to Extract Drug-Protein Relations](https://github.com/Shekswess/drugprotrelrl)  
+💻 **Code & configs:** [github.com/Shekswess/drugprotrelrl](https://github.com/Shekswess/drugprotrelrl)
+
+## Model Details
+
+| Property | Value |
+|---|---|
+| Base Model | [arcee-ai/Trinity-Mini](https://huggingface.co/arcee-ai/Trinity-Mini) |
+| Architecture | Sparse MoE (26B total / 3B active) |
+| Fine-tuning Method | LoRA (Low-Rank Adaptation) |
+| Training Method | GRPO (Reinforcement Learning) |
+| Training Data | [maziyar/OpenMed_DrugProt](https://huggingface.co/datasets/OpenMed/drugprot-parquet) |
+| Task | Drug-protein relation extraction (13-way classification) |
+| Trainable Parameters | LoRA rank=16, all projection layers |
+| License | Apache 2.0 |
+
+## Training Configuration
+
+| Parameter | Value |
+|---|---|
+| LoRA Alpha (α) | 64 |
+| LoRA Rank | 16 |
+| Target Modules | q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj + experts |
+| Learning Rate | 3e-6 |
+| Batch Size | 128 |
+| Rollouts per Example | 8 |
+| Max Generation Tokens | 2048 |
+| Temperature | 0.7 |
+
+## Quick Start
+
+**Installation**
+
+```bash
+pip install transformers peft torch accelerate
+```
+
+**Usage**
+
+```python
+from peft import PeftModel
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+
+base_model_id = "arcee-ai/Trinity-Mini"
+adapter_id = "lokahq/Trinity-Mini-DrugProt-Think"
+
+tokenizer = AutoTokenizer.from_pretrained(base_model_id)
+model = AutoModelForCausalLM.from_pretrained(
+    base_model_id,
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    trust_remote_code=True
+)
+model = PeftModel.from_pretrained(model, adapter_id)
+
+messages = [
+    {
+        "role": "user",
+        "content": (
+            "Abstract: [PASTE PUBMED ABSTRACT HERE]\n\n"
+            "Chemical entity: [DRUG NAME]\n"
+            "Protein entity: [PROTEIN NAME]\n\n"
+            "What is the relationship between the chemical and protein entities? "
+            "Choose from: A) INHIBITOR B) SUBSTRATE C) INDIRECT-DOWNREGULATOR "
+            "D) INDIRECT-UPREGULATOR E) AGONIST F) ANTAGONIST G) ACTIVATOR "
+            "H) PRODUCT-OF I) AGONIST-ACTIVATOR J) INDIRECT-UPREGULATOR "
+            "K) PART-OF L) SUBSTRATE_PRODUCT-OF M) NOT\n\n"
+            "Think step by step, then provide your answer in \\boxed{} format."
+        )
+    }
+]
+
+text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+inputs = tokenizer(text, return_tensors="pt").to(model.device)
+outputs = model.generate(**inputs, max_new_tokens=2048, temperature=0.7, top_p=0.75)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+## Training Progress
+
+Training ran for ~130 steps on Prime Intellect infrastructure. Best accuracy reward reached ~0.83 during training.
+
+| Step | Accuracy Reward | Composite Reward |
+|---|---|---|
+| 0 | ~0.68 | ~0.72 |
+| 25 | ~0.71 | ~0.74 |
+| 50 | ~0.74 | ~0.77 |
+| 75 | ~0.77 | ~0.80 |
+| 100 | ~0.80 | ~0.83 |
+
+## Limitations
+
+- This is a LoRA adapter and requires the base model ([arcee-ai/Trinity-Mini](https://huggingface.co/arcee-ai/Trinity-Mini)) to run
+- Evaluated on training-split held-out data; not yet benchmarked on the official DrugProt test set
+- Optimized specifically for 13-way DrugProt classification; may not generalize to other biomedical RE tasks
+
+## Citation
+
+```bibtex
+@article{jakimovski2026drugprotrl,
+  title   = {Post-Training an Open MoE to Extract Drug-Protein Relations},
+  author  = {Jakimovski, Bojan and Kalinovski, Petar},
+  year    = {2026},
+  url     = {https://github.com/Shekswess/drugprotrelrl}
+}
+```
+
+## Acknowledgements
+
+- [Arcee AI](https://www.arcee.ai/) for the Trinity Mini base model
+- [Prime Intellect](https://www.primeintellect.ai/) for training infrastructure
+- [maziyar](https://huggingface.co/maziyar) for the OpenMed DrugProt RL environment
+- [Hugging Face](https://huggingface.co/) for the PEFT library
+
+## Authors
+
+[Bojan Jakimovski](mailto:bojan.jakimovski@loka.com) · [Petar Kalinovski](mailto:petar.kalinovski@loka.com) · [Loka](https://loka.com)

chat_template_adapters.jinja

@@ -0,0 +1,30 @@
+You are an expert biomedical relation extraction assistant. Your task is to identify the type of interaction between a drug/chemical and a gene/protein in biomedical text.
+
+For each question:
+1. First, wrap your detailed biomedical reasoning inside <think></think> tags
+2. Analyze the context around both entities to understand their relationship
+3. Consider the pharmacological and molecular mechanisms involved
+4. Then provide your final answer inside \boxed{} using exactly one letter (A-M)
+
+The 13 DrugProt relation types are:
+A. INDIRECT-DOWNREGULATOR - Chemical indirectly decreases protein activity/expression
+B. INDIRECT-UPREGULATOR - Chemical indirectly increases protein activity/expression
+C. DIRECT-REGULATOR - Chemical directly regulates protein (mechanism unspecified)
+D. ACTIVATOR - Chemical activates the protein
+E. INHIBITOR - Chemical inhibits the protein
+F. AGONIST - Chemical acts as an agonist of the receptor/protein
+G. AGONIST-ACTIVATOR - Chemical is both agonist and activator
+H. AGONIST-INHIBITOR - Chemical is agonist but inhibits downstream effects
+I. ANTAGONIST - Chemical acts as an antagonist of the receptor/protein
+J. PRODUCT-OF - Chemical is a product of the enzyme
+K. SUBSTRATE - Chemical is a substrate of the enzyme
+L. SUBSTRATE_PRODUCT-OF - Chemical is both substrate and product
+M. PART-OF - Chemical is part of the protein complex
+
+Example format:
+<think>
+The text describes [chemical] and [protein]. Based on the context...
+- The phrase "[relevant text]" indicates that...
+- This suggests a [type] relationship because...
+</think>
+\boxed{A}