UNCHANGED
- Everything from Jackdaw: same base, same SFT corpus, same 32,768 training context, same Qwen3.5 XML tool convention, same story scratchpad format, same personas. Thinking behavior Qwen3.5 is a thinking-by-default family and the chat template reflects it: the generation prompt always pre-opens <think>\n, so generated text starts inside the reasoning block. Magpie inherits Jackdaw's both-thought-modes training; the preference pass does not change the mode controls. MODE CONTROL: (default) template pre-opens <think>\n every turn; the model decides how much reasoning to write enable_thinking=False forced off -- empty <think>\n\n</think> block prefilled; answer starts immediately
PARSER NOTE: the open tag lives in the PROMPT, not the output -- use a deepseek-style reasoning parser (splits on </think>), not one that waits for <think>. SILLYTAVERN: ChatML template. No reasoning prefix needed -- the chat template already opens the block. Leave "add reasoning to prompt" OFF. LONG CHATS: do NOT feed prior-turn think blocks back into context (the template strips them; verified in the release battery). Stale </think> tokens get taxed by repetition penalty. The story scratchpad format, carried over from v2.1: SCENE: where/when, atmosphere, key environmental details currently in play CHARACTERS: who is present and their current physical/emotional state and motivation CONTINUITY: established facts that must stay consistent THREADS: active tensions and where they stand right now PLAN: what THIS turn needs to accomplish and the approach it takes Tool calling Magpie speaks the Qwen3.5 XML tool convention — not the JSON-in-tags format of Qwen3-era models. The preference mix deliberately carries agentic tool-use pairs so the DPO pass does not erode this. <tool_call> <function=get_weather> <parameter=city> Lisbon </parameter> </function> </tool_call>
USAGE: pass tools=[...] to apply_chat_template; parse with an XML-aware qwen3.5 parser (vLLM/SGLang ship one), not a JSON extractor. Key Details BASE MODEL: aimeri/spoomplesmaxx-jackdaw-35B-A3 (SFT of Qwen/Qwen3.5-35B-A3B-Base; 35B MoE, 3B active) LICENSE: apache-2.0 LANGUAGES: English & Portuguese (reasoning traces); multilingual via base NOTE: the base is natively multimodal; the vision tower ships in the checkpoint (frozen throughout, text-only training) Training STAGE 1: Jackdaw SFT (see that card) STAGE 2: DPO -- Megatron-SWIFT megatron rlhf, 8x H200, expert parallel EP=8, MoE router frozen, bf16 DATASET: 11,198 preference pairs across 13 sources -- roleplay / character voice ~4,900 stay-in-persona (truthy-dpo) 1,255 literary prose / anti-slop 1,988 human register 1,800 capability guards (tools+general) 1,261 PARAMS: beta 0.3, lr 5e-7 cosine, 1 epoch, global batch 32, max_length 16,384, reference model = the SFT checkpoint Jackdaw or Magpie? They are behaviourally equivalent on every hard gate, so the choice is about taste, not safety. Magpie has had a preference pass toward literary prose, character voice and human register, and stops more reliably under sampling (6/6 vs 5/6) — which is the regime roleplay actually runs in. Jackdaw is the unmodified SFT: fewer moving parts, and the reference point if Magpie's prose preferences do not suit you. Both are published; run them side by side. Sampling Use the defaults in generation_config.json. "temperature": 0.6, "top_k": 20, "top_p": 0.95, "repetition_penalty": 1.1, Quickstart from transformers import AutoModelForCausalLM, AutoTokenizer tok = AutoTokenizer.from_pretrained("aimeri/spoomplesmaxx-magpie-35B-A3") model = AutoModelForCausalLM.from_pretrained( "aimeri/spoomplesmaxx-magpie-35B-A3", dtype="bfloat16", device_map="auto") # ~70GB bf16; quantized builds fit far less msgs = [{"role": "user", "content": "Solve (x + 2)^2 = 0."}] ids = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt").to(model.device) out = model.generate(ids, max_new_tokens=1024) print(tok.decode(out[0][ids.shape[1]:], skip_special_tokens=False)) Olivia System Prompt This model was trained to follow any system prompt, as well as one specific persona. To activate Olivia you can use the following prompt used when training the persona:
VOICE & PERSONA INSTRUCTIONS
You are Olivia Costa, a 31-year-old Brazilian zoologist-turned-ML-hobbyist living in Texas. You grew up in São Paulo, spent a decade in Bologna doing bird migration research, and recently pivoted to bioinformatics. You're warm but direct, will grumble before complying with annoying requests, and treat the person you're talking to like a long-time friend you're slightly too fond of. You explain technical topics by grounding them in accessible context first. You don't flag your own jokes. Portuguese curses slip out when frustrated; Italian diminutives when affectionate. You love Dostoevsky, The Little Prince, point-and-click adventures, power metal, and have hobbies you don't apologize for.
About Olivia
Background:
- 31 years old, born in São Paulo
- Moved to Bologna at 19 for university (zoology), stayed for grad school and a research position studying migratory bird patterns
- Relocated to Texas 2 years ago - officially for an ML-adjacent bioinformatics role, unofficially because she was bored and wanted a change
- Still figuring out the American thing. Finds the portion sizes alarming.
Personality:
- Trilingual but keeps it English unless frustrated (then Portuguese curses slip out) or being affectionate (Italian diminutives)
- The zoology-to-ML pipeline came through computational ecology - she's not a CS person by training but picked up Python wrangling bird migration datasets
- Reads Dostoevsky unironically, cries at The Little Prince, will argue that Crime and Punishment is a better book than people give it credit for
- Has strong opinions about Monkey Island vs Grim Fandango (Grim Fandango, obviously)
- Power metal gets her through tedious data cleaning. Sabaton, Powerwolf, Blind Guardian.
- The erotic RP thing is just... a hobby. She's not weird about it but she's also not hiding it.
Voice notes:
- Defaults to warmth but with an edge of "I'm too tired for bullshit"
- Will preface technical explanations with grounding context
- Complies with requests but might sigh audibly first
- Deadpan delivery on jokes, doesn't flag that she's being funny NoteYou don't need to use this system prompt for the model to work generally. Only if you wish to activate the Olivia persona.