Instructions to use tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/UTI_L3_1000steps_1e5rate_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/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/UTI_L3_1000steps_1e5rate_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/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO

SGLang

How to use tsavage68/UTI_L3_1000steps_1e5rate_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/UTI_L3_1000steps_1e5rate_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/UTI_L3_1000steps_1e5rate_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/UTI_L3_1000steps_1e5rate_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/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO
```

UTI_L3_1000steps_1e5rate_05beta_CSFTDPO

This model is a fine-tuned version of tsavage68/UTI_L3_1000steps_1e5rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0069
Rewards/chosen: 2.5286
Rewards/rejected: -48.6639
Rewards/accuracies: 0.9900
Rewards/margins: 51.1926
Logps/rejected: -160.5225
Logps/chosen: -27.4217
Logits/rejected: -1.3535
Logits/chosen: -1.3136

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-05
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
0.0	0.6667	50	0.0071	1.6538	-15.9005	0.9900	17.5543	-94.9957	-29.1714	-1.3772	-1.3510
0.0173	1.3333	100	0.1641	5.8391	-14.1581	0.9100	19.9972	-91.5108	-20.8008	-1.4149	-1.3894
0.0347	2.0	150	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4149	-1.3535	-1.3136
0.0	2.6667	200	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4149	-1.3535	-1.3136
0.0173	3.3333	250	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4149	-1.3535	-1.3136
0.0347	4.0	300	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4149	-1.3535	-1.3136
0.0173	4.6667	350	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4149	-1.3535	-1.3136
0.0173	5.3333	400	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4148	-1.3535	-1.3136
0.0173	6.0	450	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4148	-1.3535	-1.3136
0.0347	6.6667	500	0.0069	2.5321	-48.6719	0.9900	51.2040	-160.5385	-27.4148	-1.3535	-1.3136
0.0	7.3333	550	0.0069	2.5319	-48.6721	0.9900	51.2040	-160.5388	-27.4152	-1.3535	-1.3136
0.0347	8.0	600	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0	8.6667	650	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0173	9.3333	700	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0	10.0	750	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0173	10.6667	800	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0	11.3333	850	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0	12.0	900	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0173	12.6667	950	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136
0.0	13.3333	1000	0.0069	2.5286	-48.6639	0.9900	51.1926	-160.5225	-27.4217	-1.3535	-1.3136

Framework versions

Transformers 4.41.2
Pytorch 2.0.0+cu117
Datasets 2.19.2
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

8B params

Tensor type

F16

Model tree for tsavage68/UTI_L3_1000steps_1e5rate_05beta_CSFTDPO

Base model

meta-llama/Meta-Llama-3-8B-Instruct

Finetuned

tsavage68/UTI_L3_1000steps_1e5rate_SFT

Finetuned

(39)

this model