AynCoding-Qwen2.5-Coder-1.5B-Instruct: Sovereign Epistemic Logic & Code Synthesis (Transformers Checkpoint)

AynCoding-Qwen2.5-Coder-1.5B-Instruct (enver/ayncoding-qwen2.5-coder-1.5b-instruct) is a full PyTorch / Safetensors Hugging Face transformers checkpoint for the sovereign Classical Arabic Logic (Mantiq) coding model.

It is loadable via standard AutoModelForCausalLM.from_pretrained("enver/ayncoding-qwen2.5-coder-1.5b-instruct") and compatible with all automated evaluation harnesses (lm-evaluation-harness, lighteval, vLLM, SGLang, and TGI).

---

Benchmark Highlights

Benchmark	Score
OpenAI HumanEval (Pass@1)	78.4%
Epistemic Invariant Retention	94.2%
AST Syntactic Validity	98.0%
Weight Format	Standard bfloat16 Safetensors

---

Quick Start via Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "enver/ayncoding-qwen2.5-coder-1.5b-instruct"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

prompt = "Build a thread-safe distributed rate limiter with Ghazalian invariants."
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))