Downloads · 30 days
34
41% of all-time downloads
yashkavaiya/smollm2-135m-humana-support-lora-smoke
smollm2-135m-humana-support-lora-smoke is a text generation model from yashkavaiya. Use it when you need the model to write or continue text. It is set up for peft. The card lists the license as apache-2.0.
LoRA SFT adapter on Humana grounded synthetic support chats (public CMS/Humana docs only).
Downloads · 30 days
34
41% of all-time downloads
All-time downloads
83
Public
Repo size
3.7 MB
Likes
0
Public
Click a slice to open those files.
.json4.3 MB · 51%
From the Hugging Face model README
LoRA SFT adapter on Humana grounded synthetic support chats (public CMS/Humana docs only).
Note: This is a CPU smoke fine-tune (20 steps) to validate the training pipeline end-to-end. Not a production support model. Not for clinical, coverage, eligibility, claim, authorization, payment, or medical decisions.
HuggingFaceTB/SmolLM2-135M-Instructr=8, alpha=16, dropout 0.05)q_proj, k_proj, v_proj, o_projyashkavaiya/humana-support-slm-sft-1200c537c58d1ae922f80993657407badf5b51c1f991a285d845901cd9e6fe1409bcEducation / navigation style responses that cite public guidance and refuse determinations. Smoke checkpoint for pipeline proof only.
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base = "HuggingFaceTB/SmolLM2-135M-Instruct"
repo = "yashkavaiya/smollm2-135m-humana-support-lora-smoke"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(base, device_map="auto")
model = PeftModel.from_pretrained(model, repo)
yashkavaiya/humana-support-slm-sft-1200adapter_model.safetensors / adapter_config.json — PEFT LoRAtrain_metrics.json — training receiptsample_generations.json — smoke generationsIf you use this adapter or dataset, cite the public Humana/CMS source documents referenced in the data card.