Instructions to use tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/Transaminitis_L3_1000steps_1e7rate_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/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/Transaminitis_L3_1000steps_1e7rate_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/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO

SGLang

How to use tsavage68/Transaminitis_L3_1000steps_1e7rate_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/Transaminitis_L3_1000steps_1e7rate_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/Transaminitis_L3_1000steps_1e7rate_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/Transaminitis_L3_1000steps_1e7rate_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/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO
```

Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO

This model is a fine-tuned version of tsavage68/Transaminitis_L3_1000rate_1e7_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.1933
Rewards/chosen: 0.7492
Rewards/rejected: -1.2839
Rewards/accuracies: 0.9300
Rewards/margins: 2.0331
Logps/rejected: -31.3938
Logps/chosen: -11.0420
Logits/rejected: -1.0508
Logits/chosen: -1.0376

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: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
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
0.6925	0.2	25	0.6925	-0.0045	-0.0060	0.5500	0.0014	-18.6144	-18.5795	-1.0662	-1.0649
0.6933	0.4	50	0.6925	-0.0166	-0.0181	0.5100	0.0014	-18.7354	-18.7005	-1.0667	-1.0654
0.6915	0.6	75	0.6945	-0.0053	-0.0039	0.4600	-0.0015	-18.5932	-18.5873	-1.0676	-1.0664
0.6761	0.8	100	0.7028	-0.0339	-0.0261	0.4600	-0.0078	-18.8159	-18.8731	-1.0697	-1.0686
0.6921	1.0	125	0.6820	-0.1696	-0.1980	0.5400	0.0284	-20.5345	-20.2299	-1.0740	-1.0726
0.6815	1.2	150	0.6861	-0.0069	-0.0260	0.4600	0.0191	-18.8146	-18.6028	-1.0702	-1.0689
0.6671	1.4	175	0.6780	-0.0128	-0.0468	0.5	0.0339	-19.0222	-18.6624	-1.0734	-1.0720
0.6638	1.6	200	0.6604	0.0019	-0.0673	0.7400	0.0692	-19.2276	-18.5154	-1.0758	-1.0743
0.6595	1.8	225	0.6509	0.1265	0.0377	0.8000	0.0888	-18.1775	-17.2692	-1.0741	-1.0725
0.6367	2.0	250	0.6334	0.1771	0.0509	0.8900	0.1262	-18.0452	-16.7631	-1.0738	-1.0720
0.6195	2.2	275	0.6085	0.2361	0.0531	0.8700	0.1830	-18.0241	-16.1733	-1.0733	-1.0711
0.5823	2.4	300	0.5785	0.3043	0.0504	0.8900	0.2539	-18.0504	-15.4909	-1.0732	-1.0705
0.566	2.6	325	0.5388	0.4536	0.0969	0.8800	0.3566	-17.5852	-13.9987	-1.0723	-1.0690
0.4489	2.8	350	0.4631	0.4810	-0.0948	0.9200	0.5759	-19.5031	-13.7237	-1.0716	-1.0669
0.4492	3.0	375	0.4239	0.5170	-0.1855	0.8800	0.7025	-20.4092	-13.3640	-1.0688	-1.0634
0.3953	3.2	400	0.3679	0.5401	-0.3677	0.9100	0.9078	-22.2317	-13.1329	-1.0687	-1.0618
0.3352	3.4	425	0.3283	0.6091	-0.4726	0.9200	1.0817	-23.2806	-12.4430	-1.0668	-1.0589
0.3345	3.6	450	0.3031	0.5660	-0.6308	0.9100	1.1968	-24.8626	-12.8744	-1.0634	-1.0550
0.2808	3.8	475	0.2814	0.5999	-0.7205	0.9200	1.3204	-25.7593	-12.5348	-1.0619	-1.0530
0.295	4.0	500	0.2710	0.6397	-0.7787	0.9100	1.4184	-26.3414	-12.1370	-1.0598	-1.0503
0.1522	4.2	525	0.2397	0.6658	-0.9115	0.9200	1.5773	-27.6701	-11.8764	-1.0584	-1.0478
0.2229	4.4	550	0.2333	0.6781	-0.9833	0.9400	1.6614	-28.3878	-11.7537	-1.0576	-1.0467
0.1649	4.6	575	0.2182	0.7121	-1.0636	0.9300	1.7757	-29.1910	-11.4132	-1.0554	-1.0439
0.2244	4.8	600	0.2161	0.7092	-1.1127	0.9400	1.8219	-29.6819	-11.4421	-1.0538	-1.0421
0.1693	5.0	625	0.2022	0.7249	-1.1646	0.9300	1.8895	-30.2002	-11.2852	-1.0535	-1.0411
0.1678	5.2	650	0.1995	0.7305	-1.2049	0.9300	1.9354	-30.6042	-11.2294	-1.0522	-1.0397
0.201	5.4	675	0.2023	0.7376	-1.2194	0.9400	1.9570	-30.7488	-11.1583	-1.0513	-1.0387
0.1349	5.6	700	0.2020	0.7378	-1.2410	0.9400	1.9788	-30.9646	-11.1562	-1.0514	-1.0387
0.1701	5.8	725	0.1906	0.7477	-1.2728	0.9400	2.0205	-31.2831	-11.0576	-1.0516	-1.0386
0.2496	6.0	750	0.1936	0.7436	-1.2811	0.9400	2.0246	-31.3653	-11.0984	-1.0504	-1.0375
0.1687	6.2	775	0.1954	0.7415	-1.2850	0.9300	2.0265	-31.4046	-11.1193	-1.0510	-1.0379
0.2297	6.4	800	0.1902	0.7475	-1.2899	0.9400	2.0375	-31.4538	-11.0588	-1.0511	-1.0379
0.1449	6.6	825	0.1913	0.7469	-1.2876	0.9400	2.0345	-31.4306	-11.0649	-1.0506	-1.0375
0.1558	6.8	850	0.1915	0.7522	-1.2831	0.9400	2.0353	-31.3859	-11.0121	-1.0509	-1.0376
0.1445	7.0	875	0.1907	0.7502	-1.2855	0.9300	2.0358	-31.4099	-11.0318	-1.0508	-1.0377
0.1568	7.2	900	0.1925	0.7477	-1.2878	0.9300	2.0355	-31.4327	-11.0573	-1.0507	-1.0376
0.2344	7.4	925	0.1933	0.7486	-1.2845	0.9300	2.0331	-31.3995	-11.0484	-1.0508	-1.0377
0.2176	7.6	950	0.1933	0.7492	-1.2839	0.9300	2.0331	-31.3938	-11.0420	-1.0508	-1.0376
0.14	7.8	975	0.1933	0.7492	-1.2839	0.9300	2.0331	-31.3938	-11.0420	-1.0508	-1.0376
0.1992	8.0	1000	0.1933	0.7492	-1.2839	0.9300	2.0331	-31.3938	-11.0420	-1.0508	-1.0376

Framework versions

Transformers 4.40.2
Pytorch 2.0.0+cu117
Datasets 2.19.1
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

8B params

Tensor type

F16

Model tree for tsavage68/Transaminitis_L3_1000steps_1e7rate_01beta_CSFTDPO

Base model

meta-llama/Meta-Llama-3-8B-Instruct

Finetuned

tsavage68/Transaminitis_L3_1000rate_1e7_SFT

Finetuned

(17)

this model