synoema-coder-3b-tools-v8/README.md

---
language: [en]
license: apache-2.0
base_model: unsloth/Qwen2.5-Coder-3B-Instruct
library_name: peft
pipeline_tag: text-generation
tags: [synoema, tool-use, agentic, function-calling, lora, qlora, code, qwen2.5, gguf]
model-index:
- name: synoema-coder-3b-tools-v8
  results:
  - task: {type: tool-use, name: Synoema MCP Agentic Tool-Use Eval (28 tasks)}
    metrics:
    - {type: pass@1, value: 1.0, name: "28-task agentic eval (28/28)"}
---

# Synoema-Coder-3B Tools (C8)

A **3B** LoRA fine-tune of `unsloth/Qwen2.5-Coder-3B-Instruct` that turns it into an **agentic coding model for the
[Synoema](https://synoema.tech) programming language** — it writes Synoema, type-checks it,
runs it, searches a corpus, and self-corrects on errors, all through MCP tools.

- 🌐 **Website:** https://synoema.tech
- 🤖 **This model:** https://huggingface.co/delimitter/synoema-coder-3b-tools-v8
- 📚 **Training corpus (dataset):** https://huggingface.co/datasets/delimitter/synoema-coder-3b-tools-corpus

---

## 🏆 Result: **100% (28/28)** on the Synoema agentic tool-use benchmark

Scored on the **corrected agentic harness**: the model is driven **turn-by-turn** (generation
stops at `<|im_end|>`), and **real** tool results are injected between turns — actual
`sno check` / `sno run` output from the live Synoema compiler, never mocked. A task only passes
if the model genuinely completes it end-to-end (e.g. multi-write self-correction: write broken
code → observe the type error → rewrite a valid fix → type-check passes).

| Capability | Tasks | Pass |
|---|---|---|
| Write + typecheck + run | TU1–TU3, TU5, TU10 | ✅ |
| Search → write → run | TU6, TU9, TU20 | ✅ |
| Multi-write self-correction (if/else → ternary) | TU4, TU13 | ✅ |
| Language features (ADT, HOF, pattern match, cons) | TU11, TU14–TU19, TU23, TU29 | ✅ |
| List comprehensions | TU12, TU26 | ✅ |
| Nested ternary (fizzbuzz) | TU22, TU30 | ✅ |
| **Total** | **28** | **28/28** |

---

## What is Synoema?

[Synoema](https://synoema.tech) is an **LLM-native programming language and runtime** designed so
that models can write it reliably:

- **BPE-aligned operators** — every operator maps to exactly one `cl100k_base` token.
- **Ternary instead of if/else** — `? cond -> a : b` (nestable).
- **GBNF grammar** for constrained decoding (structural-correctness guarantee).
- **Cranelift JIT + WebAssembly** compile targets.
- **MCP server** exposing `file_write`, `file_read`, `sno_typecheck`, `sno_run`, `search_corpus`.
- **Contract annotations** (`requires` / `ensures`) for formal verification.

---

## Model details

| Property | Value |
|---|---|
| Base model | `unsloth/Qwen2.5-Coder-3B-Instruct` |
| Parameters | 3B |
| Method | QLoRA (4-bit NF4 + LoRA), merged to fp16 for GGUF |
| LoRA | r=16, alpha=32 |
| Sequence length | 1024 |
| Epochs / cycle | 3 |
| Training corpus | ~18k tool-use + codegen examples — **every example passes `sno check` + `sno run`** |
| Cycle | C8 (sequential "carousel": each cycle warm-starts from the best previous adapter, then trains on the corpus plus targeted examples for the prior cycle's failures) |
| Hardware | AMD RX 7900 GRE 16GB (ROCm + unsloth) |

---

## GGUF files (llama.cpp / Ollama / LM Studio)

| File | Quant | Size | Notes |
|---|---|---|---|
| `synoema-coder-3b-tools-v8.Q4_K_M.gguf` | Q4_K_M | 2 GB | smallest, recommended for local use |
| `synoema-coder-3b-tools-v8.Q8_0.gguf` | Q8_0 | 3 GB | near-lossless |
| `synoema-coder-3b-tools-v8.f16.gguf` | F16 | 6 GB | full precision |

```bash
# llama.cpp
llama-cli -hf delimitter/synoema-coder-3b-tools-v8 --hf-file synoema-coder-3b-tools-v8.Q4_K_M.gguf -p "Write quicksort in Synoema to src/qs.sno and run it."

# Ollama
ollama run hf.co/delimitter/synoema-coder-3b-tools-v8:Q4_K_M
```

---

## Usage — Transformers + PEFT (adapter)

```python
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base = AutoModelForCausalLM.from_pretrained("unsloth/Qwen2.5-Coder-3B-Instruct", device_map="auto")
tok  = AutoTokenizer.from_pretrained("unsloth/Qwen2.5-Coder-3B-Instruct")
model = PeftModel.from_pretrained(base, "delimitter/synoema-coder-3b-tools-v8")
```

Prompt format is **ChatML**. The system prompt used at training/eval:

```
<|im_start|>system
You are sno-code, a Synoema coding agent. Use tools to write and verify code.<|im_end|>
<|im_start|>user
Write `square x = x * x` with `main = square 9` to src/square.sno, typecheck and run it.<|im_end|>
<|im_start|>assistant
```

The model emits OpenAI-style `tool_calls` for `file_write`, `sno_typecheck`, `sno_run`,
`file_read`, `search_corpus`; feed real tool results back as `tool` turns.

---

## Synoema language quick reference

```synoema
maxOf x y = ? x > y -> x : y            -- ternary (NO if/then/else)
fact 0 = 1                              -- pattern matching
fact n = n * fact (n - 1)
evens xs = [x | x <- xs, x % 2 == 0]    -- list comprehension
sumList xs = foldl (\acc x -> acc + x) 0 xs   -- higher-order functions
Direction = North | South | East | West       -- ADT
opposite North = South
main = qsort [3 1 4 1 5]                -- lists are SPACE-separated
```

---

## License

Apache-2.0 (same as the Qwen2.5 base model). **Synoema** © Andrey Bubnov — https://synoema.tech