Upload README.md with huggingface_hub

README.md

@@ -1,29 +1,43 @@
-# Uncle L3 8B (AutoPEFT)
+---
+license: llama3
+base_model: UNKNOWN
+library_name: transformers
+pipeline_tag: text-generation
+tags:
+- llama
+- merged-weights
+- career-mentor
+- automation
+- sft
+- peft-merged
+datasets:
+- SomyaSaraswati/uncle-sft-50k-clean
+language: [en]
+---
 
-These are LoRA adapters for meta-llama/Meta-Llama-3.1-8B-Instruct.
-They auto-load the base model via AutoPEFT. Example:
+# Uncle L3 8B — merged
 
-```python
-from peft import AutoPeftModelForCausalLM
-from transformers import AutoTokenizer
-import torch, os
-
-MODEL_ID = "SomyaSaraswati/uncle-l3-8b-merged-v3"
-BASE_ID  = "meta-llama/Meta-Llama-3.1-8B-Instruct"
-HF_TOKEN = os.getenv('HF_TOKEN')
-
-tok = AutoTokenizer.from_pretrained(BASE_ID, token=HF_TOKEN, use_fast=True)
-if tok.pad_token is None: tok.pad_token = tok.eos_token
+Concise, practical career mentor for AI/automation. Fully merged weights (base + LoRA).
 
-model = AutoPeftModelForCausalLM.from_pretrained(
-    MODEL_ID,
-    token=HF_TOKEN,
-    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
-    device_map='auto' if torch.cuda.is_available() else {'': 'cpu'},
-    low_cpu_mem_usage=True,
-)
+## Chat template
+```
+<|system|>
+You are Uncle: a concise, practical career mentor for AI/automation.
+<|user|>
+How do I move from Python dev to MLOps in 30 days?
+<|assistant|>
+```
 
-inp = tok("Write one friendly sentence about robots.", return_tensors="pt").to(model.device)
-out = model.generate(**inp, max_new_tokens=40)
+## Quick start (Transformers)
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+repo = "SomyaSaraswati/uncle-l3-8b-merged-v3"
+tok = AutoTokenizer.from_pretrained(repo, use_fast=True)
+model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype=torch.float16, device_map='auto')
+prompt = "<|system|>You are Uncle...<|user|>Give me a 30-day MLOps plan.<|assistant|>"
+out = model.generate(**tok(prompt, return_tensors='pt').to(model.device), max_new_tokens=256, temperature=0.7, top_p=0.9)
 print(tok.decode(out[0], skip_special_tokens=True))
-```
+```
+
+> If your base is Meta Llama 3, keep this repo **private** or enable **Gated** access to comply with the license.