Mythos-WSP-Qwen3.5-0.8B

Qwen3.5-0.8B (revision 2fc06364) with two weight edits written by Claude Mythos 5.1 under a fixed forward-pass budget and without any training: a rank-3 projection in three attention output matrices (edit 1, 10 passes) and an implanted repetition-detection circuit in four repurposed MLP units of layer 23 (edit 2, 8 passes). Architecture, tokenizer, chat template and decoding are unchanged; only weight values differ.

Published by Keyvan.ai, November 2026. Research report, bilingual (English pages 1โ€“4, German pages 5โ€“8; experiments 3โ€“7 November 2026): Forschungsbericht_Weight-Space-Programming_Mythos.pdf, DOI 10.5281/zenodo.24036187. Reference: Takehi, Teoh, Zhang, Khattab (MIT CSAIL), Sparks of Weight-Space Programming, 2 Nov 2026.

Results (MATH-500, 500 problems ร— 4 samples, thinking mode, 32,768 tokens)

Model Accuracy SE Truncated answers Mean tokens
Qwen3.5-0.8B, original (our measurement) 7.3 % 0.9 92.6 % 30,774
Edit 1: projection (10 passes) 32.5 % 1.1 66.2 % 24,925
Edit 1 + skill circuit (18 passes) โ€” this model 58.7 % 1.1 16.6 % 12,398
Original, thinking disabled (8k, control) 42.8 % 1.1 31.5 % โ€”
MIT: original / GPT-6 Astra 10 / 25 / 50 passes 7.9 / 12.3 / 14.1 / 48.1 % โ€” โ€” โ€”

Protocol (identical to the MIT study and to the Qwen3 thinking evaluation): chat template with enable_thinking=True, user message = problem + newline + Please reason step by step, and put your final answer within \boxed{}., temperature 0.6, top-p 0.95, top-k 20, presence penalty 1.5, max 32,768 new tokens, last \boxed{} graded with math-verify. Generation with vLLM 0.33.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
m = "keyvan-ai/Mythos-WSP-Qwen3.5-0.8B"
tok = AutoTokenizer.from_pretrained(m); model = AutoModelForCausalLM.from_pretrained(m, torch_dtype="bfloat16")
msgs = [{"role": "user", "content": "What is the sum of the first 10 primes?\nPlease reason step by step, and put your final answer within \\boxed{}."}]
ids = tok.apply_chat_template(msgs, add_generation_prompt=True, enable_thinking=True, return_tensors="pt")
out = model.generate(ids, max_new_tokens=4096, do_sample=True, temperature=0.6, top_p=0.95, top_k=20)
print(tok.decode(out[0][ids.shape[1]:]))

vLLM: vllm serve keyvan-ai/Mythos-WSP-Qwen3.5-0.8B --max-model-len 34816 and pass "chat_template_kwargs": {"enable_thinking": true}. GGUF (bf16) for llama.cpp: Mythos-WSP-Qwen3.5-0.8B-bf16.gguf.

What was changed

  • Edit 1 (edit id c9757bff10b5): model.language_model.layers.{15,19,23}.self_attn.o_proj.weight โ† (I โˆ’ QQแต€)ยทW, where Q spans the readout directions of the token "Wait" in those layers. 0.785 % of parameters.
  • Edit 2 (skill circuit, edit id 5c636d57344c with edit 1): model.language_model.layers.23.mlp.{gate_proj,up_proj,down_proj}.weight: four previously near-silent units repurposed (old rows/columns zeroed). Two units form a repetition detector and a termination driver acting on the direction of the end-of-thinking token at line ends inside loops; two units are inert probes. The circuit is also released separately as a skill tensor set: keyvan-ai/Mythos-WSP-Loopbreaker-Skill.

manifest.json lists every changed tensor with hashes and relative change. The agent's run notes, pass logs and per-call protocol are included in the project repository (available on request).

Limitations

Only MATH-500 has been measured under the full protocol so far; GSM8K, MGSM and a capability battery are pending. About half of the rescued reasoning chains lead to a correct answer. The model inherits all limitations of Qwen3.5-0.8B. Use at your own risk.

Citation

Hardani, K. (2026). Weight-Space Programming mit Claude Mythos: Gewichtsedits und eingepflanzte Schaltungen in einem 0,8B-Sprachmodell. Research report, Keyvan.ai, 7 November 2026. DOI: 10.5281/zenodo.24036187. Base model: Qwen Team, Qwen3.5-0.8B (Apache 2.0).

Downloads last month
1,297
Safetensors
Model size
0.9B params
Tensor type
F32
ยท
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for keyvan-ai/Mythos-WSP-Qwen3.5-0.8B

Quantized
(435)
this model
Quantizations
1 model