Instructions to use tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT")
model = AutoModelForCausalLM.from_pretrained("tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT", 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_600_STEPS_05beta_5e7rate_CDPOSFT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT"
# 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_600_STEPS_05beta_5e7rate_CDPOSFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT

SGLang

How to use tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT 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_600_STEPS_05beta_5e7rate_CDPOSFT" \
    --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_600_STEPS_05beta_5e7rate_CDPOSFT",
		"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_600_STEPS_05beta_5e7rate_CDPOSFT" \
        --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_600_STEPS_05beta_5e7rate_CDPOSFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT with Docker Model Runner:
```
docker model run hf.co/tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT
```

chat_600_STEPS_05beta_5e7rate_CDPOSFT

This model is a fine-tuned version of tsavage68/chat_600STEPS_1e8rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6669
Rewards/chosen: -0.0665
Rewards/rejected: -0.1611
Rewards/accuracies: 0.5275
Rewards/margins: 0.0946
Logps/rejected: -19.1242
Logps/chosen: -16.8876
Logits/rejected: -0.5967
Logits/chosen: -0.5966

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: 5e-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: 600

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6903	0.0977	50	0.6936	0.0166	0.0155	0.4000	0.0011	-18.7710	-16.7214	-0.5983	-0.5982
0.6671	0.1953	100	0.6792	-0.0508	-0.0879	0.4835	0.0371	-18.9777	-16.8562	-0.6007	-0.6006
0.6942	0.2930	150	0.6855	-0.1406	-0.1792	0.4791	0.0386	-19.1604	-17.0359	-0.5997	-0.5996
0.6826	0.3906	200	0.6802	-0.0490	-0.1057	0.4835	0.0567	-19.0134	-16.8527	-0.5953	-0.5952
0.7074	0.4883	250	0.6747	-0.0391	-0.1111	0.4967	0.0721	-19.0242	-16.8328	-0.5930	-0.5929
0.6745	0.5859	300	0.6694	-0.0467	-0.1352	0.5011	0.0885	-19.0723	-16.8480	-0.5980	-0.5979
0.6636	0.6836	350	0.6685	-0.0796	-0.1700	0.5253	0.0905	-19.1420	-16.9137	-0.5947	-0.5945
0.6607	0.7812	400	0.6691	-0.0747	-0.1648	0.5209	0.0902	-19.1317	-16.9040	-0.5986	-0.5984
0.6758	0.8789	450	0.6693	-0.0676	-0.1582	0.5275	0.0906	-19.1183	-16.8898	-0.5967	-0.5965
0.6562	0.9766	500	0.6686	-0.0674	-0.1598	0.5187	0.0924	-19.1216	-16.8894	-0.5965	-0.5964
0.5185	1.0742	550	0.6689	-0.0681	-0.1596	0.5077	0.0915	-19.1213	-16.8909	-0.5971	-0.5970
0.5392	1.1719	600	0.6669	-0.0665	-0.1611	0.5275	0.0946	-19.1242	-16.8876	-0.5967	-0.5966

Framework versions

Transformers 4.40.1
Pytorch 2.0.0+cu117
Datasets 2.19.1
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/chat_600_STEPS_05beta_5e7rate_CDPOSFT

Base model

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

Finetuned

tsavage68/chat_600STEPS_1e8rate_SFT

Finetuned

(23)

this model