Instructions to use tsavage68/chat_1000STEPS_1e6_05beta_DPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/chat_1000STEPS_1e6_05beta_DPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/chat_1000STEPS_1e6_05beta_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_1e6_05beta_DPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/chat_1000STEPS_1e6_05beta_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_1e6_05beta_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_1e6_05beta_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_1e6_05beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/chat_1000STEPS_1e6_05beta_DPO

SGLang

How to use tsavage68/chat_1000STEPS_1e6_05beta_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_1e6_05beta_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_1e6_05beta_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_1e6_05beta_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_1e6_05beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/chat_1000STEPS_1e6_05beta_DPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/chat_1000STEPS_1e6_05beta_DPO
```

chat_1000STEPS_1e6_05beta_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: 0.7047
Rewards/chosen: -0.5484
Rewards/rejected: -0.8442
Rewards/accuracies: 0.5319
Rewards/margins: 0.2958
Logps/rejected: -20.4796
Logps/chosen: -17.8414
Logits/rejected: -0.6334
Logits/chosen: -0.6333

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: 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
0.6923	0.2	100	0.6978	-0.3692	-0.4056	0.4549	0.0364	-19.6025	-17.4830	-0.6054	-0.6052
0.7106	0.39	200	0.7053	0.1136	-0.0026	0.4791	0.1161	-18.7964	-16.5175	-0.6058	-0.6056
0.5991	0.59	300	0.7229	-0.2199	-0.3741	0.4879	0.1541	-19.5394	-17.1845	-0.6117	-0.6115
0.7082	0.78	400	0.7221	-0.0056	-0.1904	0.5033	0.1848	-19.1721	-16.7559	-0.5870	-0.5868
0.6684	0.98	500	0.7010	-0.1029	-0.3043	0.5275	0.2014	-19.3998	-16.9504	-0.5454	-0.5452
0.2004	1.17	600	0.6974	-0.4104	-0.6928	0.5341	0.2824	-20.1768	-17.5654	-0.6005	-0.6004
0.2715	1.37	700	0.7012	-0.5147	-0.8128	0.5429	0.2981	-20.4169	-17.7741	-0.6258	-0.6257
0.2303	1.56	800	0.7031	-0.5366	-0.8347	0.5341	0.2981	-20.4606	-17.8177	-0.6321	-0.6320
0.2729	1.76	900	0.7052	-0.5480	-0.8437	0.5341	0.2957	-20.4787	-17.8406	-0.6333	-0.6331
0.2621	1.95	1000	0.7047	-0.5484	-0.8442	0.5319	0.2958	-20.4796	-17.8414	-0.6334	-0.6333

Framework versions

Transformers 4.37.2
Pytorch 2.0.0+cu117
Datasets 2.17.0
Tokenizers 0.15.2

Downloads last month: 5

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/chat_1000STEPS_1e6_05beta_DPO

Base model

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

Finetuned

(552)

this model