Add model card

README.md

@@ -0,0 +1,62 @@
+---
+language: en
+license: mit
+base_model: microsoft/Phi-3.5-mini-instruct
+tags:
+  - project-management
+  - communication
+  - lora
+  - peft
+  - phi-3.5
+pipeline_tag: text-generation
+---
+
+# PMCommunicator
+
+PMCommunicator is a LoRA fine-tune of [Phi-3.5-mini-instruct](https://huggingface.co/microsoft/Phi-3.5-mini-instruct)
+(3.8B parameters) specialized for generating professional project management communications.
+
+Given a project context (from PMPlanner + PMReasoner), it generates stakeholder-ready prose:
+kickoff emails, status reports, risk escalation memos, executive summaries, board updates,
+and project closeout reports.
+
+## Model Details
+
+| Property | Value |
+|---|---|
+| Base model | microsoft/Phi-3.5-mini-instruct (3.8B) |
+| Fine-tuning method | LoRA (PEFT) |
+| LoRA rank | 16, alpha 32 |
+| Trainable params | 25M / 3.82B (0.65%) |
+| Training data | 28,000+ PM communication examples |
+| Val loss | 0.0105 |
+| License | MIT |
+
+## Usage
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+model_id = "pmcore/pmcommunicator"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, device_map="auto")
+
+system = (
+    "You are PMCommunicator, an expert project manager and communications specialist. "
+    "Generate professional, stakeholder-ready project communications based on the provided "
+    "project context. Be specific — use the actual project name, numbers, and timeline. "
+    "Write in clear business English. Output only the communication document itself."
+)
+user = "Project: Cloud migration, 50 legacy apps, 18 months, $8M budget. 3 phases planned.\n\nWrite a weekly status report for stakeholders."
+
+prompt = f"<|system|>\n{system}<|end|>\n<|user|>\n{user}<|end|>\n<|assistant|>\n"
+inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.3, do_sample=True)
+print(tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))
+```
+
+## Full Pipeline
+
+Use PMCommunicator as part of the full PMCore pipeline for best results.
+See [PMCore on GitHub](https://github.com/snavazio/pmcore).