troy-cli 0.2.0__tar.gz → 0.2.1__tar.gz

{troy_cli-0.2.0 → troy_cli-0.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: troy-cli
-Version: 0.2.0
+Version: 0.2.1
 Summary: Fine-tune LLMs on your MacBook with one YAML file. Built for Apple Silicon.
 Author: Troy
 License: Apache-2.0
@@ -46,8 +46,10 @@ troy doctor            # check your Mac: chip, memory, MLX, what you can train
 troy init              # create troy.yaml + sample data
 troy train             # fine-tune (LoRA/QLoRA via MLX)
 troy chat              # talk to the result
+troy eval              # did it work? base-vs-tuned loss + samples
 troy serve             # OpenAI-compatible API at localhost:8080/v1
 troy export -f gguf    # ship it to llama.cpp / Ollama / LM Studio
+troy push you/model    # upload to the Hugging Face Hub
 ```
 
 ## The config is the interface
@@ -75,7 +77,7 @@ output: ./output
 
 | Command | Purpose |
 |---|---|
-| `troy init` | Create a config from a template (`chat`, `dpo`) |
+| `troy init` | Create a config from a template (`chat`, `dpo`, `orpo`) |
 | `troy doctor` | Hardware + dependency check, with model-size guidance |
 | `troy train` | LoRA fine-tuning: SFT, DPO, or ORPO |
 | `troy chat` | Interactive REPL (or `-p` for one-shot) with your adapter |

{troy_cli-0.2.0 → troy_cli-0.2.1}/README.md

@@ -31,8 +31,10 @@ troy doctor            # check your Mac: chip, memory, MLX, what you can train
 troy init              # create troy.yaml + sample data
 troy train             # fine-tune (LoRA/QLoRA via MLX)
 troy chat              # talk to the result
+troy eval              # did it work? base-vs-tuned loss + samples
 troy serve             # OpenAI-compatible API at localhost:8080/v1
 troy export -f gguf    # ship it to llama.cpp / Ollama / LM Studio
+troy push you/model    # upload to the Hugging Face Hub
 ```
 
 ## The config is the interface
@@ -60,7 +62,7 @@ output: ./output
 
 | Command | Purpose |
 |---|---|
-| `troy init` | Create a config from a template (`chat`, `dpo`) |
+| `troy init` | Create a config from a template (`chat`, `dpo`, `orpo`) |
 | `troy doctor` | Hardware + dependency check, with model-size guidance |
 | `troy train` | LoRA fine-tuning: SFT, DPO, or ORPO |
 | `troy chat` | Interactive REPL (or `-p` for one-shot) with your adapter |

{troy_cli-0.2.0 → troy_cli-0.2.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "troy-cli"
-version = "0.2.0"
+version = "0.2.1"
 description = "Fine-tune LLMs on your MacBook with one YAML file. Built for Apple Silicon."
 readme = "README.md"
 requires-python = ">=3.10"

{troy_cli-0.2.0 → troy_cli-0.2.1}/src/troy/__init__.py

@@ -1,3 +1,3 @@
 """Troy: fine-tune LLMs on your MacBook with one YAML file."""
 
-__version__ = "0.2.0"
+__version__ = "0.2.1"

{troy_cli-0.2.0 → troy_cli-0.2.1}/src/troy/cli.py

@@ -51,7 +51,7 @@ def _main(
 @app.command()
 def init(
     template: str = typer.Option(
-        "chat", help="Template: chat (SFT) or dpo (preference tuning)."
+        "chat", help="Template: chat (SFT), dpo, or orpo (preference tuning)."
     ),
     path: Path = typer.Option(Path("troy.yaml"), help="Where to write the config."),
     force: bool = typer.Option(False, "--force", help="Overwrite an existing config."),

{troy_cli-0.2.0 → troy_cli-0.2.1}/src/troy/templates.py

@@ -50,6 +50,32 @@ training:
 output: ./output
 """
 
+ORPO_TEMPLATE = """\
+# Troy config — preference tuning (ORPO: no reference model, ~2x DPO speed)
+# Run with: troy train
+
+base: mlx-community/Qwen3-0.6B-4bit
+task: orpo
+
+data:
+  train: ./data/preferences.jsonl   # {"prompt": ..., "chosen": ..., "rejected": ...}
+  format: preference
+  val_split: 0.05
+
+training:
+  epochs: 1
+  lr: 2e-5
+  batch_size: auto
+  seq_len: 2048
+  lora:
+    r: 8
+    alpha: 16
+  orpo:
+    lambda: 0.2
+
+output: ./output
+"""
+
 SAMPLE_SFT_DATA = [
     {
         "instruction": "What is the capital of France?",
@@ -81,4 +107,5 @@ SAMPLE_DPO_DATA = [
 TEMPLATES = {
     "chat": (CHAT_TEMPLATE, "train.jsonl", SAMPLE_SFT_DATA),
     "dpo": (DPO_TEMPLATE, "preferences.jsonl", SAMPLE_DPO_DATA),
+    "orpo": (ORPO_TEMPLATE, "preferences.jsonl", SAMPLE_DPO_DATA),
 }