Instructions to use tsavage68/chat_1000STEPS_1e7rate_01beta_DPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/chat_1000STEPS_1e7rate_01beta_DPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/chat_1000STEPS_1e7rate_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_1e7rate_01beta_DPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/chat_1000STEPS_1e7rate_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_1e7rate_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_1e7rate_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_1e7rate_01beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/chat_1000STEPS_1e7rate_01beta_DPO

SGLang

How to use tsavage68/chat_1000STEPS_1e7rate_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_1e7rate_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_1e7rate_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_1e7rate_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_1e7rate_01beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/chat_1000STEPS_1e7rate_01beta_DPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/chat_1000STEPS_1e7rate_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: 0.6919
Rewards/chosen: -0.0000
Rewards/rejected: -0.0027
Rewards/accuracies: 0.4637
Rewards/margins: 0.0027
Logps/rejected: -18.8181
Logps/chosen: -16.7447
Logits/rejected: -0.5977
Logits/chosen: -0.5976

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-07
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	0.6935	-0.0018	-0.0012	0.4044	-0.0006	-18.8033	-16.7622	-0.5978	-0.5977
0.6937	0.39	200	0.6928	-0.0003	-0.0010	0.4505	0.0007	-18.8010	-16.7472	-0.5978	-0.5977
0.6901	0.59	300	0.6923	-0.0008	-0.0025	0.4527	0.0018	-18.8166	-16.7523	-0.5969	-0.5968
0.6912	0.78	400	0.6922	0.0001	-0.0020	0.4549	0.0020	-18.8109	-16.7440	-0.5982	-0.5981
0.6912	0.98	500	0.6922	0.0001	-0.0020	0.4813	0.0020	-18.8108	-16.7437	-0.5979	-0.5978
0.689	1.17	600	0.6920	-0.0008	-0.0033	0.4637	0.0025	-18.8240	-16.7525	-0.5979	-0.5978
0.6898	1.37	700	0.6916	0.0003	-0.0029	0.5055	0.0032	-18.8205	-16.7416	-0.5979	-0.5977
0.6876	1.56	800	0.6921	-0.0011	-0.0033	0.4593	0.0022	-18.8246	-16.7559	-0.5981	-0.5979
0.6902	1.76	900	0.6917	-0.0000	-0.0030	0.4637	0.0030	-18.8217	-16.7450	-0.5974	-0.5973
0.6883	1.95	1000	0.6919	-0.0000	-0.0027	0.4637	0.0027	-18.8181	-16.7447	-0.5977	-0.5976

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_1e7rate_01beta_DPO

Base model

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

Finetuned

(552)

this model