Instructions to use tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_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/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_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/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO

SGLang

How to use tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_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/Summary4500_M2_1000steps_1e5rate_01beta_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/Summary4500_M2_1000steps_1e5rate_01beta_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/Summary4500_M2_1000steps_1e5rate_01beta_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/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO
```

Hyponatremia_M2_1000steps_1e5rate_01beta_CSFTDPO

This model is a fine-tuned version of tsavage68/Summary4500_M2_200steps_1e7rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0014
Rewards/chosen: -17.0154
Rewards/rejected: -135.8507
Rewards/accuracies: 0.9980
Rewards/margins: 118.8353
Logps/rejected: -1511.2372
Logps/chosen: -263.8938
Logits/rejected: -2.3218
Logits/chosen: -2.3112

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: 1
eval_batch_size: 1
seed: 42
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

Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.0112	50	0.0031	-11.3626	-30.1735	0.9980	18.8108	-454.4643	-207.3659	-1.9304	-1.9034
0.0224	100	0.0046	-55.0573	-119.5048	0.9960	64.4476	-1347.7778	-644.3123	-2.1229	-2.0968
0.0336	150	0.0268	-125.6898	-283.0312	0.9920	157.3413	-2983.0410	-1350.6383	-0.7037	-0.6736
0.0448	200	0.0268	-125.6898	-283.0312	0.9920	157.3413	-2983.0410	-1350.6383	-0.7037	-0.6736
0.0559	250	0.1221	-118.0507	-242.0082	0.9900	123.9576	-2572.8118	-1274.2462	-2.0982	-2.0597
0.0671	300	0.1221	-118.0507	-242.0082	0.9900	123.9576	-2572.8118	-1274.2462	-2.0982	-2.0597
0.0783	350	0.0014	-139.8033	-344.3079	0.9980	204.5046	-3595.8086	-1491.7727	-1.8323	-1.8165
0.0895	400	0.0014	-17.4091	-78.1513	0.9980	60.7421	-934.2426	-267.8311	-2.3714	-2.3343
0.1007	450	0.0014	-17.4091	-78.1513	0.9980	60.7421	-934.2426	-267.8311	-2.3714	-2.3343
0.1119	500	0.0014	-17.4091	-78.1513	0.9980	60.7421	-934.2426	-267.8311	-2.3714	-2.3343
0.1231	550	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.1343	600	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.1454	650	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.1566	700	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.1678	750	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.1790	800	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.1902	850	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.2014	900	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.2126	950	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112
0.2238	1000	0.0014	-17.0154	-135.8507	0.9980	118.8353	-1511.2372	-263.8938	-2.3218	-2.3112

Framework versions

Transformers 4.42.4
Pytorch 2.0.0+cu117
Datasets 2.20.0
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/Summary4500_M2_1000steps_1e5rate_01beta_CSFTDPO

Base model

mistralai/Mistral-7B-Instruct-v0.2

Finetuned

tsavage68/Summary4500_M2_200steps_1e7rate_SFT

Finetuned

(14)

this model