Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.
Python cliconsumer-gpudpofine-tuningggufhuggingfacellmllmopslocal-ailocal-llmloralow-vrammachine-learningollamapeftpythonpytorchqlorasfttransformers
docs(contributors): #2051 [skip ci]
Alpamysc1fab5bf6541
diff --git a/CONTRIBUTORS.md b/CONTRIBUTORS.mdindex 3b1cd9a0..a1f2d359 100644--- a/CONTRIBUTORS.md+++ b/CONTRIBUTORS.md@@ -892,6 +892,7 @@ Listed by first contribution. PR numbers link the work.- `training.auto_mixed_precision` loaded on all 23 tasks, but only the SFT trainer (and `tts` through it) reads it, so a DPO, GRPO or KTO run that set it trained at its trainer's default precision and said nothing, and Autopilot's flat `generate_config` helper wrote the key for any task. They refused `true` outside a named `AMP_APPLYING_TASKS = {sft, tts}` at config load, with the field, the task and the fix in the message. `false` still loads everywhere. The validator sits between the #795 resolver and QuEST check and the unsloth refusal, the generator is scoped to the same set, and the GRPO `grpo_fp16` message, the field description and the docs were corrected. **The part worth copying is the test table**: the gate is parametrized over every value of the `task` Literal with `sft`/`tts` listed literally, so a new task is refused by default and the controls cannot drift with the constant. When review found that nothing pinned SFT keeping the field in the generator, they added that control. 7329 tests pass across 61 related files (23 of their tests fail on `main`); 8 mutations killed in the first round and 6 in the final one ([#1618](https://github.com/MakazhanAlpamys/Soup/issues/1618), [#1627](https://github.com/MakazhanAlpamys/Soup/pull/1627))- Every `SOUP_LAYER_STREAM_STRIPE_DIRS` refusal printed the entry with `repr`, so a Windows path came out with every backslash doubled and could not be copied back. They added `quote_path()`, which keeps a path as typed inside single quotes and escapes exactly the control characters the validation refuses, and used it at every site in the issue's table, including four Windows API errors that only fire on Windows. **They listed the spellings first (drive letter, UNC, spaces, trailing separator, POSIX, relative) and tested each one.** After review they added the changelog fragment, moved the helper to a neutral module and added a UNC case. 22 new tests (the file does not collect on `main` without the helper), 11 of 11 mutations killed ([#1611](https://github.com/MakazhanAlpamys/Soup/issues/1611), [#1626](https://github.com/MakazhanAlpamys/Soup/pull/1626))- Improved loopback Web UI host validation and IPv6 controls ([#1829](https://github.com/MakazhanAlpamys/Soup/issues/1829), [#1839](https://github.com/MakazhanAlpamys/Soup/pull/1839))+ - `backend: mlx` accepted `data.chat_template` and never applied it, so a prefix-stable override changed nothing and a Qwen3.5 run still died in the response-mask check after the model had loaded. They confirmed on `main` that the same config trains on the transformers backend, then refused the combination at config load with a message that names the field, the backend and both ways out, parametrised the tests over two registered names and a raw Jinja string on sft, dpo and orpo with the empty-string and unsloth controls, and removed the MLX DPO/ORPO warning and registry rows the refusal made unreachable ([#2075](https://github.com/MakazhanAlpamys/Soup/issues/2075), [#2051](https://github.com/MakazhanAlpamys/Soup/pull/2051))- **Akshat Shah** ([@akshatshahh](https://github.com/akshatshahh))- First contribution: `http://[::1]:<port>` was accepted by the judge API-base and vLLM checks but refused by eval-gate suites, `soup ship --judge-model` and the online-DPO trainer, because three gates kept their own `("localhost", "127.0.0.1")` literal and two more kept a third copy. They made all five import `net_guard.LOOPBACK_HOSTS` and added an AST ratchet that fails on any rebuilt loopback literal, including single-quoted and multi-line spellings. Their 27 tests: 11 fail on main, 27 pass here; 7 mutations, all killed ([#1548](https://github.com/MakazhanAlpamys/Soup/issues/1548), [#1554](https://github.com/MakazhanAlpamys/Soup/pull/1554))