Instructions to use tsavage68/chat_1000STEPS_1e5rate_01beta_DPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/chat_1000STEPS_1e5rate_01beta_DPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/chat_1000STEPS_1e5rate_01beta_DPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/chat_1000STEPS_1e5rate_01beta_DPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/chat_1000STEPS_1e5rate_01beta_DPO", 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/chat_1000STEPS_1e5rate_01beta_DPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/chat_1000STEPS_1e5rate_01beta_DPO"
# 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/chat_1000STEPS_1e5rate_01beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/chat_1000STEPS_1e5rate_01beta_DPO

SGLang

How to use tsavage68/chat_1000STEPS_1e5rate_01beta_DPO 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/chat_1000STEPS_1e5rate_01beta_DPO" \
    --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/chat_1000STEPS_1e5rate_01beta_DPO",
		"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/chat_1000STEPS_1e5rate_01beta_DPO" \
        --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/chat_1000STEPS_1e5rate_01beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/chat_1000STEPS_1e5rate_01beta_DPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/chat_1000STEPS_1e5rate_01beta_DPO
```

chat_1000STEPS_1e7rate_01beta_DPO

This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.9688
Rewards/chosen: -2.8329
Rewards/rejected: -3.3687
Rewards/accuracies: 0.4989
Rewards/margins: 0.5358
Logps/rejected: -52.4786
Logps/chosen: -45.0740
Logits/rejected: -0.2885
Logits/chosen: -0.2875

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: 4
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 8
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
0.8129	0.2	100	0.7825	-1.1957	-1.1981	0.3934	0.0024	-30.7728	-28.7020	-0.0569	-0.0566
0.8136	0.39	200	0.8828	-0.9245	-0.8916	0.4044	-0.0329	-27.7071	-25.9900	0.2762	0.2769
0.7535	0.59	300	0.8597	-1.3930	-1.4515	0.4000	0.0585	-33.3058	-30.6746	1.0803	1.0813
0.9558	0.78	400	0.8896	-0.8319	-0.7033	0.3604	-0.1285	-25.8247	-25.0635	0.4421	0.4425
0.7839	0.98	500	0.7987	-0.8948	-1.0616	0.4264	0.1667	-29.4069	-25.6928	0.6877	0.6886
0.2401	1.17	600	0.9002	-2.8266	-3.2238	0.4725	0.3972	-51.0296	-45.0107	-0.0174	-0.0164
0.2852	1.37	700	0.9362	-2.6553	-3.0787	0.4769	0.4234	-49.5784	-43.2978	-0.1079	-0.1069
0.2151	1.56	800	0.9663	-2.5826	-3.1268	0.5011	0.5443	-50.0594	-42.5702	-0.1730	-0.1719
0.2376	1.76	900	0.9701	-2.8346	-3.3672	0.4945	0.5326	-52.4633	-45.0905	-0.2881	-0.2870
0.2943	1.95	1000	0.9688	-2.8329	-3.3687	0.4989	0.5358	-52.4786	-45.0740	-0.2885	-0.2875

Framework versions

Transformers 4.37.2
Pytorch 2.0.0+cu117
Datasets 2.17.0
Tokenizers 0.15.2

Downloads last month: 6

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/chat_1000STEPS_1e5rate_01beta_DPO

Base model

meta-llama/Llama-2-7b-chat-hf

Finetuned

(552)

this model