GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
ValiantLabs
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
ValiantLabs
Model Tree
Qwen/Qwen3-14B
Input Modalities
Output Modalities
Supported Functionality
Guardpoint delivers structured medical responses using the Qwen 3 prompt format.
Guardpoint is a reasoning finetune; we recommend enable_thinking=True for all chats.
Example inference script to get started:
python
from transformers import AutoModelForCausalLM, AutoTokenizer model_name = "ValiantLabs/Qwen3-14B-Guardpoint" # load the tokenizer and the modeltokenizer = AutoTokenizer.from_pretrained(model_name)model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto") # prepare the model inputprompt = "A 60-year-old undergoes a Total Knee Arthroplasty (TKA). Post-operatively, they complain of a clunking sensation and instability when descending stairs. On exam, they have excessive posterior translation of the tibia at 90 degrees of flexion. The TKA used a Cruciate Retaining (CR) implant. Diagnosis is PCL incompetence or rupture. Explain why a CR implant relies on a functional PCL for femoral rollback and how converting to a Posterior Stabilized (PS) implant resolves this biomechanical failure."#prompt = "I have that tube in my chest for dialysis while my arm heals. The dressing came off in the shower and the tube got tugged a bit. It didn't come out, but now there's this red cuff thing showing that used to be inside the skin. It’s sticking out about an inch. Can I just push it back in and tape it?"#prompt = "In the workup of a tumor of unknown primary, a biopsy shows a poorly differentiated carcinoma. The IHC profile is: CK7+, CK20+, CDX2+, TTF-1 negative, PAX8 negative. Based on this cytokeratin and transcription factor profile, where is the most likely primary site of the malignancy?"#prompt = "I have bad arthritis in my lower back and hips. I saw a chiropractor who said my 'pelvis is twisted' and wants to do high-velocity adjustments. My rheumatologist said absolutely not because of my 'osteophytes'. Who is right? I just want to walk without stiffness."messages = [ {"role": "user", "content": prompt}]text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True, enable_thinking=True # Switches between thinking and non-thinking modes. Default is True.)model_inputs = tokenizer([text], return_tensors="pt").to(model.device) # conduct text completiongenerated_ids = model.generate( **model_inputs, max_new_tokens=32768)output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() # parsing thinking contenttry: # rindex finding 151668 (</think>) index = len(output_ids) - output_ids[::-1].index(151668)except ValueError: index = 0 thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n") print("thinking content:", thinking_content)print("content:", content)DISCLAIMER: Guardpoint is a medical reasoning finetune that is subject to the strengths and weaknesses of LLMs. A conversation with an LLM is not a substitute for a professional medical examination. Utilize Guardpoint responsibly.
Guardpoint is created by Valiant Labs.
Check out our HuggingFace page to see Shining Valiant, Esper, and all of our models!
We care about open source. For everyone to use.
from transformers import AutoModelForCausalLM, AutoTokenizer model_name = "ValiantLabs/Qwen3-14B-Guardpoint" # load the tokenizer and the modeltokenizer = AutoTokenizer.from_pretrained(model_name)model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto") # prepare the model inputprompt = "A 60-year-old undergoes a Total Knee Arthroplasty (TKA). Post-operatively, they complain of a clunking sensation and instability when descending stairs. On exam, they have excessive posterior translation of the tibia at 90 degrees of flexion. The TKA used a Cruciate Retaining (CR) implant. Diagnosis is PCL incompetence or rupture. Explain why a CR implant relies on a functional PCL for femoral rollback and how converting to a Posterior Stabilized (PS) implant resolves this biomechanical failure."#prompt = "I have that tube in my chest for dialysis while my arm heals. The dressing came off in the shower and the tube got tugged a bit. It didn't come out, but now there's this red cuff thing showing that used to be inside the skin. It’s sticking out about an inch. Can I just push it back in and tape it?"#prompt = "In the workup of a tumor of unknown primary, a biopsy shows a poorly differentiated carcinoma. The IHC profile is: CK7+, CK20+, CDX2+, TTF-1 negative, PAX8 negative. Based on this cytokeratin and transcription factor profile, where is the most likely primary site of the malignancy?"#prompt = "I have bad arthritis in my lower back and hips. I saw a chiropractor who said my 'pelvis is twisted' and wants to do high-velocity adjustments. My rheumatologist said absolutely not because of my 'osteophytes'. Who is right? I just want to walk without stiffness."messages = [ {"role": "user", "content": prompt}]text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True, enable_thinking=True # Switches between thinking and non-thinking modes. Default is True.)model_inputs = tokenizer([text], return_tensors="pt").to(model.device) # conduct text completiongenerated_ids = model.generate( **model_inputs, max_new_tokens=32768)output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() # parsing thinking contenttry: # rindex finding 151668 (</think>) index = len(output_ids) - output_ids[::-1].index(151668)except ValueError: index = 0 thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n") print("thinking content:", thinking_content)print("content:", content)