Instructions to use tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO

SGLang

How to use tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO 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 "tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO" \
    --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": "tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO",
		"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 "tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO" \
        --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": "tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO
```

UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO

This model is a fine-tuned version of tsavage68/UTI_M2_1000steps_1e7rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.5476
Rewards/chosen: 0.9941
Rewards/rejected: -7.4997
Rewards/accuracies: 0.2100
Rewards/margins: 8.4937
Logps/rejected: -24.3733
Logps/chosen: -2.5543
Logits/rejected: -2.7955
Logits/chosen: -2.7946

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-06
train_batch_size: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.5541	0.3333	25	0.5519	0.0709	-2.5498	0.2100	2.6207	-14.4735	-4.4006	-2.6521	-2.6515
0.5545	0.6667	50	0.8852	-0.2566	-3.1766	0.1800	2.9200	-15.7271	-5.0556	-2.6377	-2.6371
0.5718	1.0	75	0.7287	-0.0078	-6.5265	0.2000	6.5188	-22.4270	-4.5580	-2.5987	-2.5978
2.2289	1.3333	100	0.5476	0.0080	-5.0654	0.2100	5.0734	-19.5048	-4.5265	-2.6769	-2.6758
0.5545	1.6667	125	0.5476	-0.0935	-5.6463	0.2100	5.5528	-20.6666	-4.7295	-2.6679	-2.6668
0.5545	2.0	150	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5545	2.3333	175	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.4852	2.6667	200	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.6412	3.0	225	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5545	3.3333	250	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5372	3.6667	275	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5892	4.0	300	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.4679	4.3333	325	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5718	4.6667	350	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5199	5.0	375	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5025	5.3333	400	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5199	5.6667	425	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5025	6.0	450	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5718	6.3333	475	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5718	6.6667	500	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5025	7.0	525	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5199	7.3333	550	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5372	7.6667	575	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5718	8.0	600	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5372	8.3333	625	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.4332	8.6667	650	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5372	9.0	675	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5025	9.3333	700	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5025	9.6667	725	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5892	10.0	750	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5199	10.3333	775	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5199	10.6667	800	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5372	11.0	825	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5199	11.3333	850	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.6065	11.6667	875	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5718	12.0	900	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.4159	12.3333	925	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.6238	12.6667	950	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.6065	13.0	975	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946
0.5025	13.3333	1000	0.5476	0.9941	-7.4997	0.2100	8.4937	-24.3733	-2.5543	-2.7955	-2.7946

Framework versions

Transformers 4.41.2
Pytorch 2.0.0+cu117
Datasets 2.19.2
Tokenizers 0.19.1

Downloads last month: 5

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/UTI2_M2_1000steps_1e6rate_05beta_CSFTDPO

Base model

mistralai/Mistral-7B-Instruct-v0.2

Finetuned

tsavage68/UTI_M2_1000steps_1e7rate_SFT

Finetuned

(32)

this model