Instructions to use alterego238/Qwen-2.5-7B-Simple-RL-debug with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use alterego238/Qwen-2.5-7B-Simple-RL-debug with Transformers:

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

pipe = pipeline("text-generation", model="alterego238/Qwen-2.5-7B-Simple-RL-debug")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("alterego238/Qwen-2.5-7B-Simple-RL-debug")
model = AutoModelForCausalLM.from_pretrained("alterego238/Qwen-2.5-7B-Simple-RL-debug", 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 alterego238/Qwen-2.5-7B-Simple-RL-debug with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "alterego238/Qwen-2.5-7B-Simple-RL-debug"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "alterego238/Qwen-2.5-7B-Simple-RL-debug",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/alterego238/Qwen-2.5-7B-Simple-RL-debug

SGLang

How to use alterego238/Qwen-2.5-7B-Simple-RL-debug 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 "alterego238/Qwen-2.5-7B-Simple-RL-debug" \
    --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": "alterego238/Qwen-2.5-7B-Simple-RL-debug",
		"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 "alterego238/Qwen-2.5-7B-Simple-RL-debug" \
        --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": "alterego238/Qwen-2.5-7B-Simple-RL-debug",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use alterego238/Qwen-2.5-7B-Simple-RL-debug with Docker Model Runner:
```
docker model run hf.co/alterego238/Qwen-2.5-7B-Simple-RL-debug
```

Qwen-2.5-7B-Simple-RL-debug / trainer_state.json

alterego238

Model save

841231f verified over 1 year ago

Raw

History Blame Contribute Delete

2.42 kB

	{
	"best_metric": null,
	"best_model_checkpoint": null,
	"epoch": 1.0,
	"eval_steps": 8,
	"global_step": 8,
	"is_hyper_param_search": false,
	"is_local_process_zero": true,
	"is_world_process_zero": true,
	"log_history": [
	{
	"clip_ratio": 0.0,
	"completion_length": 866.75,
	"epoch": 0.125,
	"grad_norm": 0.0,
	"kl": 0.0,
	"learning_rate": 3e-06,
	"loss": 0.0,
	"reward": 0.0,
	"reward_std": 0.0,
	"rewards/accuracy_reward": 0.0,
	"rewards/format_reward": 0.0,
	"step": 1
	},
	{
	"clip_ratio": 0.0,
	"completion_length": 449.25,
	"epoch": 0.625,
	"grad_norm": 0.043186482042074203,
	"kl": 0.0009279251098632812,
	"learning_rate": 1.1662185990655286e-06,
	"loss": 0.0,
	"reward": 0.8125,
	"reward_std": 0.26933756470680237,
	"rewards/accuracy_reward": 0.8125,
	"rewards/format_reward": 0.0,
	"step": 5
	},
	{
	"epoch": 1.0,
	"eval_clip_ratio": 0.0,
	"eval_completion_length": 570.34375,
	"eval_kl": 0.0012311935424804688,
	"eval_loss": 4.939456994179636e-05,
	"eval_reward": 0.78125,
	"eval_reward_std": 0.27900634706020355,
	"eval_rewards/accuracy_reward": 0.78125,
	"eval_rewards/format_reward": 0.0,
	"eval_runtime": 78.5017,
	"eval_samples_per_second": 0.102,
	"eval_steps_per_second": 0.025,
	"step": 8
	},
	{
	"clip_ratio": 0.0,
	"completion_length": 647.6666666666666,
	"epoch": 1.0,
	"kl": 0.0002209345499674479,
	"reward": 0.9166666666666666,
	"reward_std": 0.16666666666666666,
	"rewards/accuracy_reward": 0.9166666666666666,
	"rewards/format_reward": 0.0,
	"step": 8,
	"total_flos": 0.0,
	"train_loss": -0.062465423718094826,
	"train_runtime": 165.9317,
	"train_samples_per_second": 0.048,
	"train_steps_per_second": 0.048
	}
	],
	"logging_steps": 5,
	"max_steps": 8,
	"num_input_tokens_seen": 0,
	"num_train_epochs": 1,
	"save_steps": 500,
	"stateful_callbacks": {
	"TrainerControl": {
	"args": {
	"should_epoch_stop": false,
	"should_evaluate": false,
	"should_log": false,
	"should_save": false,
	"should_training_stop": false
	},
	"attributes": {}
	}
	},
	"total_flos": 0.0,
	"train_batch_size": 1,
	"trial_name": null,
	"trial_params": null
	}