This repository contains a PEFT LoRA adapter for Qwen/Qwen3-32B, released as Polyhymnia in the Presence Muses series.
It is the full-data supervised fine-tune (SFT) of the series for multi-turn roleplay / companion chat with send_message tool calls.
The base model is not included. Load the adapter on top of
Qwen/Qwen3-32B.
| Item | Value |
|---|---|
| Developed by | PresenceAI |
| Series | Presence Muses |
| Release | v1.0 (GA) |
| Model type | LoRA adapter (PEFT) |
| Training stage | Supervised fine-tuning (SFT) |
| Base model | Qwen/Qwen3-32B |
| Languages | English, Chinese |
| License | Apache-2.0 |
| Adapter rank | r = 32, alpha = 32, dropout = 0.0 |
| Target modules | q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj |
Qwen/Qwen3-32B ──SFT (full data)──▶ Polyhymnia-32B-SFT-LoRA-v1.0 (this repository)
Trained directly from the base model. No intermediate checkpoint or upstream adapter is required — load this adapter on Qwen/Qwen3-32B as shown below. The base weights are not included in this repository.
send_message call.Out of scope: factual question answering, medical / legal / financial advice, high-stakes decision making.
Automatic evaluation on an internal held-out split (token-level eval loss):
| Step | Eval loss | Note |
|---|---|---|
| 4,200 | 0.6023 | Best value recorded in the training log |
| 4,300 | 0.6029 | |
| 4,500 | 0.6058 |
The training run continued beyond the recorded evaluation history. The value above is the best recorded during training and was not re-measured on the exact released checkpoint. No task-level benchmark, human preference study, or safety evaluation is included.
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
base_id = "Qwen/Qwen3-32B"
adapter_id = "PresenceAI/Polyhymnia-32B-SFT-LoRA-v1.0"
tokenizer = AutoTokenizer.from_pretrained(base_id)
model = AutoModelForCausalLM.from_pretrained(
base_id, torch_dtype="bfloat16", device_map="auto"
)
model = PeftModel.from_pretrained(model, adapter_id)
model.eval()
vllm serve Qwen/Qwen3-32B \
--enable-lora --max-lora-rank 32 \
--lora-modules polyhymnia=PresenceAI/Polyhymnia-32B-SFT-LoRA-v1.0 \
--enable-auto-tool-choice --tool-call-parser hermes
Tool-call format used during training:
<tool_call>
{"name": "send_message", "arguments": {"text": "..."}}
</tool_call>
Use the tokenizer and chat_template.jinja included in this repository.
The adapter is released under the Apache License 2.0. The base model Qwen/Qwen3-32B is also Apache-2.0. See LICENSE and NOTICE.
@misc{presenceai_polyhymnia_32b_sft_lora_v1,
title = {PresenceAI Polyhymnia-32B-SFT-LoRA v1.0: roleplay and tool-calling chat},
author = {PresenceAI},
year = {2026},
note = {LoRA adapter, Apache-2.0}
}
Qwen/Qwen3-32B