Instructions to use gemechisw/Tenacious-DPO-LoRA-v0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use gemechisw/Tenacious-DPO-LoRA-v0.1 with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("unsloth/Qwen2.5-3B-Instruct-bnb-4bit")
model = PeftModel.from_pretrained(base_model, "gemechisw/Tenacious-DPO-LoRA-v0.1")

Transformers

How to use gemechisw/Tenacious-DPO-LoRA-v0.1 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="gemechisw/Tenacious-DPO-LoRA-v0.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("gemechisw/Tenacious-DPO-LoRA-v0.1", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use gemechisw/Tenacious-DPO-LoRA-v0.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "gemechisw/Tenacious-DPO-LoRA-v0.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "gemechisw/Tenacious-DPO-LoRA-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/gemechisw/Tenacious-DPO-LoRA-v0.1

SGLang

How to use gemechisw/Tenacious-DPO-LoRA-v0.1 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "gemechisw/Tenacious-DPO-LoRA-v0.1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "gemechisw/Tenacious-DPO-LoRA-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "gemechisw/Tenacious-DPO-LoRA-v0.1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "gemechisw/Tenacious-DPO-LoRA-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Desktop
Docker Model Runner
How to use gemechisw/Tenacious-DPO-LoRA-v0.1 with Docker Model Runner:
```
docker model run hf.co/gemechisw/Tenacious-DPO-LoRA-v0.1
```

gemechisw commited on May 2

Commit

0a7bace

verified ·

1 Parent(s): fe2b0b0

Updated readme, removed slate links

Browse files

Files changed (1) hide show

README.md +3 -6

README.md CHANGED Viewed

@@ -22,18 +22,15 @@ PEFT LoRA adapter trained with DPO for Tenacious-Bench sales-agent intervention
 This artifact is an adapter-only checkpoint (not a full merged foundation model). It was trained on preference pairs derived from Tenacious-Bench v0.1 train/dev splits, where chosen outputs pass benchmark constraints and rejected outputs represent policy/reliability failures.
 - **Developed by:** Gemechis Worku
-- **Funded by [optional]:** [More Information Needed]
 - **Shared by [optional]:** Gemechis Worku
 - **Model type:** PEFT LoRA adapter trained with DPO (Path B)
 - **Language(s) (NLP):** English
-- **License:** [More Information Needed]
 - **Finetuned from model [optional]:** `unsloth/Qwen2.5-3B-Instruct-bnb-4bit`
 ### Model Sources [optional]
 - **Repository:** https://github.com/gemechisworku/tenacious_bench_v01
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
 ## Uses
@@ -98,7 +95,7 @@ Dataset reference:
 Preprocessing and controls:
 1. Held-out split excluded from preference construction.
 2. Chosen/rejected pairs derived under benchmark rubric constraints.
-3. Leakage controls documented in `methodology_rationale.md` and implementation plan ACT III.
 ### Training Procedure
@@ -128,7 +125,7 @@ Preprocessing and controls:
 1. Train runtime: `994.5271` seconds (~16.6 minutes).
 2. Train samples/sec: `0.251`.
 3. Train steps/sec: `0.032`.
-4. Reported train loss: `0.2612`.
 5. Artifacts tracked in:
    - `training/config.yaml`
    - `training/metrics.json`

 This artifact is an adapter-only checkpoint (not a full merged foundation model). It was trained on preference pairs derived from Tenacious-Bench v0.1 train/dev splits, where chosen outputs pass benchmark constraints and rejected outputs represent policy/reliability failures.
 - **Developed by:** Gemechis Worku
 - **Shared by [optional]:** Gemechis Worku
 - **Model type:** PEFT LoRA adapter trained with DPO (Path B)
 - **Language(s) (NLP):** English
+- **License:** MIT
 - **Finetuned from model [optional]:** `unsloth/Qwen2.5-3B-Instruct-bnb-4bit`
 ### Model Sources [optional]
 - **Repository:** https://github.com/gemechisworku/tenacious_bench_v01
 ## Uses
 Preprocessing and controls:
 1. Held-out split excluded from preference construction.
 2. Chosen/rejected pairs derived under benchmark rubric constraints.
+3. Leakage controls documented in `methodology_rationale.md`
 ### Training Procedure
 1. Train runtime: `994.5271` seconds (~16.6 minutes).
 2. Train samples/sec: `0.251`.
 3. Train steps/sec: `0.032`.
+4. Reported train loss: `0.03`.
 5. Artifacts tracked in:
    - `training/config.yaml`
    - `training/metrics.json`