Upload README.md with huggingface_hub

README.md

@@ -2,26 +2,55 @@
 license: apache-2.0
 base_model: Qwen/Qwen2.5-Coder-1.5B-Instruct
 pipeline_tag: text-generation
-tags: [code, python, qwen2.5-coder, dora, reasoning]
+library_name: transformers
+tags: [code, python, qwen2.5-coder, dora, mixture-of-models, code-generation]
 language: [en]
 ---
 
-# Python SLM 1.5B — v3 (r=64, reasoning) — RECOMMENDED
+# MoM-Python-SLM (1.5B)
 
-Python-specialized fine-tune of **Qwen2.5-Coder-1.5B-Instruct** for the Mixture-of-Models (MoM) mesh.
-Single-turn code **generator** (not an agent).
+The **Python code-generation node** of a **Mixture-of-Models (MoM)** mesh — a set of small,
+specialized Qwen2.5-Coder SLMs (shared tokenizer) coordinated by a lightweight router, aiming to beat
+frontier generalists on coding by *specialization depth* rather than parameter count.
 
-- **Base:** Qwen/Qwen2.5-Coder-1.5B-Instruct
-- **Method:** DoRA r=64 (4.6% trainable), SFT on reasoning-augmented Python data.
-- **Result (native harness, greedy pass@1):** **HumanEval 70.7% / MBPP 69.6%** — beats base
-  68.9%/66.7% (delta +1.8 / +2.9). First checkpoint to beat its all-language base at Python.
-- See also `python-slm-v4` (98% reasoning coverage, experimental).
+This node is a **single-turn code generator** (not an agent): given a Python task (optionally with an
+upstream context packet), it returns reasoning followed by code. It shares the Qwen2.5-Coder
+tokenizer with the other generative nodes, which is what makes logit-space fusion across the mesh
+valid.
+
+- **Base:** [Qwen/Qwen2.5-Coder-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-Coder-1.5B-Instruct)
+- **Method:** DoRA r=64 (≈4.6% trainable), SFT (Phase A 1ep + Phase B 2ep), then merged.
+- **Data:** 476K instances (decontaminated vs HumanEval/MBPP, 0 overlap) built from the complete
+  CPython docs + Flask/Requests source, issues/PRs, CVEs, and execution-verified synthetic problems.
+
+## Benchmarks (greedy pass@1)
+
+| Suite | Metric | base | **this model** |
+|---|---|---|---|
+| HumanEval | pass@1 | 68.9 | **70.7** |
+| MBPP | pass@1 | 66.7 | **69.6** |
+| Domain (held-out) | `spec_to_code` exec | 0.632 | **0.714** (+8.2) |
+| Domain (held-out) | `api_signature` param-recall | 0.217 | **0.299** (+8.2) |
+| Domain (held-out) | `problem_solving` exec | 0.700 | 0.713 (parity) |
+
+The largest gains are on **library/API capability** (writing correct code from a spec, recalling API
+signatures) — the dimension HumanEval/MBPP are saturated on and can't measure. The repo's
+self-contained domain-eval notebook reproduces these.
+
+## Recipe findings (load-bearing)
+- **Low DoRA rank wins:** r=64 specializes without forgetting; r=256 catastrophically regressed
+  (HumanEval 60.4 < base).
+- **Moderate reasoning wins:** the ~25%-reasoning recipe (this model) beat a 98%-reasoning sibling,
+  whose HumanEval *collapsed* to 47 (always-reason prose fights the signature-completion format).
 
 ## Usage
 ```python
 from transformers import AutoModelForCausalLM, AutoTokenizer
-tok = AutoTokenizer.from_pretrained("srivarenya/python-slm-v3")
-model = AutoModelForCausalLM.from_pretrained("srivarenya/python-slm-v3", torch_dtype="bfloat16", device_map="auto")
+tok = AutoTokenizer.from_pretrained("srivarenya/MoM-python-slm")
+model = AutoModelForCausalLM.from_pretrained(
+    "srivarenya/MoM-python-slm", dtype="bfloat16", device_map="auto")
 ```
+Prompt with the training system prompt + a Python task; the model returns reasoning then code.
 
-Code, recipe, eval harness: https://github.com/srivarenya01/python-slm
+Next step in the pipeline: **GRPO/RLVR** against an execution-grounded reward to push past the
+instruct-tuning ceiling. Code, training recipe, and eval harnesses: project repository.