GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
adastracomputing
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
adastracomputing
Model Tree
Qwen/Qwen2.5-Coder-32B-Instruct
Input Modalities
Output Modalities
Supported Functionality
Given a CVE and a nixpkgs package, the model acts as an agent that:
read_nixpkgs_file)find_upstream_fix)prefetch_hash)submit_for_approval)The model uses Qwen 2.5's native <tool_call> / <tool_response> XML format for tool interactions.
| Parameter | Value |
|---|---|
| Base model | Qwen/Qwen2.5-Coder-32B-Instruct (4-bit quantized) |
| Method | QLoRA (SFT) |
| LoRA rank | 8 |
| LoRA alpha | 16 |
| Target modules | q_proj, k_proj, v_proj, o_proj |
| Adapter size | 64 MB |
| Training examples | 588 train / 66 eval |
| Epochs | 3 (108 steps) |
| Effective batch size | 16 (batch 1 x grad accum 16) |
| Learning rate | 1e-4 (cosine schedule) |
| Max sequence length | 4096 |
| Precision | bf16 |
| Hardware | NVIDIA L40S (48 GB) |
| Training time | ~2.3 hours |
| Step | Train Loss | Eval Loss | Token Accuracy |
|---|---|---|---|
| 5 | 1.569 | -- | 70.0% |
| 50 | 0.758 | 0.625 | -- |
| 100 | 0.517 | 0.472 | -- |
| 108 | 0.511 | -- | -- |
| Version | Rank | Adapter Size | Notes |
|---|---|---|---|
| v9 (current) | 8 | 64 MB | Reduced rank for faster cold starts on Workers AI |
| v8 | 32 | 256 MB | Cold-start timeouts on Workers AI due to large adapter |
| v4-v7 | 32 | 256 MB | Failed: model_type: "qwen2" rejected by CF |
Trained on adastracomputing/nixpkgs-security-patches -- 654 multi-turn tool-calling conversations derived from real merged security PRs in NixOS/nixpkgs.
This adapter is designed for Cloudflare Workers AI deployment. Key requirements:
model_type must be "llama" in adapter_config.json (CF rejects "qwen2", but "llama" works due to similar architecture)text
stop: ["</tool_call>", "<tool_response>", "<|im_end|>", "<|endoftext|>"]adapter_config.json + adapter_model.safetensors (max 2 assets)bash
npx wrangler ai finetune create @cf/qwen/qwen2.5-coder-32b-instruct <name> <adapter_folder>/python
from peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizer base = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen2.5-Coder-32B-Instruct", device_map="auto", torch_dtype="auto",)model = PeftModel.from_pretrained(base, "adastracomputing/nixpkgs-security-qwen-lora")tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-Coder-32B-Instruct").nix files would likely improve results.Apache 2.0
npx wrangler ai finetune create @cf/qwen/qwen2.5-coder-32b-instruct <name> <adapter_folder>/from peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizer base = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen2.5-Coder-32B-Instruct", device_map="auto", torch_dtype="auto",)model = PeftModel.from_pretrained(base, "adastracomputing/nixpkgs-security-qwen-lora")tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-Coder-32B-Instruct")