Instructions to use mjf-su/New-Edge-Critic with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mjf-su/New-Edge-Critic with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-VL-4B-Instruct")
model = PeftModel.from_pretrained(base_model, "mjf-su/New-Edge-Critic")

Transformers

How to use mjf-su/New-Edge-Critic with Transformers:

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

pipe = pipeline("text-generation", model="mjf-su/New-Edge-Critic")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("mjf-su/New-Edge-Critic", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use mjf-su/New-Edge-Critic with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/mjf-su/New-Edge-Critic

SGLang

How to use mjf-su/New-Edge-Critic 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 "mjf-su/New-Edge-Critic" \
    --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": "mjf-su/New-Edge-Critic",
		"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 "mjf-su/New-Edge-Critic" \
        --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": "mjf-su/New-Edge-Critic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use mjf-su/New-Edge-Critic with Docker Model Runner:
```
docker model run hf.co/mjf-su/New-Edge-Critic
```

mjf-su commited on May 23

Commit

d968a11

verified ·

1 Parent(s): 9b64c90

Upload train_config.json with huggingface_hub

Browse files

Files changed (1) hide show

train_config.json +34 -0

train_config.json ADDED Viewed

	@@ -0,0 +1,34 @@

+{
+  "train_dataset": "datasets/Faithfulness-Critic-Dataset/train_dataset.jsonl",
+  "val_dataset": "datasets/Faithfulness-Critic-Dataset/val_dataset.jsonl",
+  "dataset_dir": "datasets/Faithfulness-Critic-Dataset",
+  "max_length": 4096,
+  "seed": 0,
+  "eval_max_new_tokens": 32,
+  "eval_batch_size": 8,
+  "eval_steps": 531,
+  "scene_oversample_factor": 1.8,
+  "e1_oversample_factor": 3.2,
+  "base_model": "Qwen/Qwen3-VL-4B-Instruct",
+  "use_lora": true,
+  "lora_rank": 256,
+  "lora_alpha": null,
+  "lora_dropout": 0.05,
+  "lora_target_modules": "q_proj,k_proj,v_proj,o_proj",
+  "flash_attn": false,
+  "gradient_checkpointing": false,
+  "output_dir": "results/scene-critic-lora",
+  "epochs": 2.0,
+  "per_device_batch_size": 8,
+  "grad_accum_steps": 1,
+  "lr": 0.0001,
+  "warmup_ratio": 0.03,
+  "weight_decay": 0.0,
+  "logging_steps": 10,
+  "save_steps": 2656,
+  "save_total_limit": 2,
+  "max_steps": -1,
+  "report_to": "wandb",
+  "wandb_project": "Faithfulness-Critic",
+  "wandb_run_name": "completion-verdict"
+}