GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
richowell7
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
richowell7
Model Tree
HuggingFaceTB/SmolLM2-135M-Instruct
Input Modalities
Output Modalities
Supported Functionality
gpt-6-luna (respuestas de máximo 90 palabras), en 6 temas.gpt-6-luna) contra la respuesta de referencia.| Antes (modelo base) | Después (destilado) | |
|---|---|---|
| Nota media del juez (1–10) | 2.00 | 2.75 |
| Aprobación (nota ≥ 7) | 0% | 0% |
| Tema | Antes | Después |
|---|---|---|
| Agentes de IA | 2.67 | 2.33 |
| Ajuste fino | 2.00 | 2.50 |
| Fundamentos de LLM | 1.50 | 2.50 |
| Ingeniería de prompts | 2.00 | 3.33 |
| RAG | 1.50 | 3.00 |
Cifras calculadas desde salidas/examen_antes.jsonl y salidas/examen_despues.jsonl. El juez es otro LLM: su nota tiene ruido
(≈ ±0,3 entre corridas) y, en una auditoría manual de 10 ítems, resultó levemente más generoso que un humano.
python
from transformers import AutoModelForCausalLM, AutoTokenizerfrom peft import PeftModel base = "HuggingFaceTB/SmolLM2-135M-Instruct"tok = AutoTokenizer.from_pretrained(base)model = PeftModel.from_pretrained(AutoModelForCausalLM.from_pretrained(base), "richowell7/taller-destilacion-slm") msgs = [{"role": "system", "content": "You are an expert assistant in generative AI. Answer in English, clearly and briefly."}, {"role": "user", "content": "When should a company use RAG instead of fine-tuning?"}]enc = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt", return_dict=True)out = model.generate(**enc, max_new_tokens=120, do_sample=False, repetition_penalty=1.15)print(tok.decode(out[0, enc["input_ids"].shape[1]:], skip_special_tokens=True))from transformers import AutoModelForCausalLM, AutoTokenizerfrom peft import PeftModel base = "HuggingFaceTB/SmolLM2-135M-Instruct"tok = AutoTokenizer.from_pretrained(base)model = PeftModel.from_pretrained(AutoModelForCausalLM.from_pretrained(base), "richowell7/taller-destilacion-slm") msgs = [{"role": "system", "content": "You are an expert assistant in generative AI. Answer in English, clearly and briefly."}, {"role": "user", "content": "When should a company use RAG instead of fine-tuning?"}]enc = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt", return_dict=True)out = model.generate(**enc, max_new_tokens=120, do_sample=False, repetition_penalty=1.15)print(tok.decode(out[0, enc["input_ids"].shape[1]:], skip_special_tokens=True))