Downloads · 30 days
46
13% of all-time downloads
Danny-Dasilva/Qwen3-14B-antidoom-W4A16-GPTQ-g32-DSpark
Qwen3-14B-antidoom-W4A16-GPTQ-g32-DSpark is a text generation model from Danny-Dasilva. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
An Antidoom (FTPO anti-repetition) version of Qwen/Qwen3-14B, quantized post-training to GPTQ int4 (W4A16, group=32, symmetric) and verified as a DSpark speculative-decoding target (draft: deepseek-ai/dsparkqwen314bbl…
Downloads · 30 days
46
13% of all-time downloads
All-time downloads
353
Public
Parameters
15.2B
10.6 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors10.5 GB · 100%
How the weights are stored.
I3213.2B · 87%
From the Hugging Face model README
An Antidoom (FTPO anti-repetition) version of
Qwen/Qwen3-14B, quantized post-training to
GPTQ int4 (W4A16, group=32, symmetric) and verified as a
DSpark speculative-decoding target
(draft: deepseek-ai/dspark_qwen3_14b_block7).
263 tok/s single-stream on one RTX 5090 (32 GB) with DSpark k=7 + CUDA graphs (1.82× over native) — at 8.4 GB of weights.
chosen_win ≥ 0.4 (hit at epoch ~3.9).
The anti-loop preference flips from 5% to 45% chosen-win during training.DSpark acceptance is preserved: 25.5% (original, official AWQ reference) → 30.8% (this model, antidoom + GPTQ g32). The FTPO patch is local and targeted; it does not break the draft head.
| config | tok/s | DSpark accept |
|---|---|---|
| this model + DSpark k=7 + CUDA graphs | 263 (peaks 343) | 30.8% |
| this model, native (no speculation) | 144 | — |
| reference: official Qwen3-14B-AWQ + DSpark k=7 | 244 | 25.5% |
| reference: official Qwen3-14B-AWQ, native | 147 | — |
2,028 held-out prompts per arm (training prompts excluded), temperature 0.01, up to 4k tokens, loop detection via the antidoom pipeline's own detector:
| model | doom-loop rate |
|---|---|
| original Qwen3-14B (official AWQ) | 21/2039 = 1.03% |
| this model (antidoom) | 8/2039 = 0.39% |
~60% relative reduction in runaway-repetition incidence (two-proportion z ≈ 2.3, p ≈ 0.02), from only 58 FTPO pairs — measured, not assumed.
from vllm import LLM, SamplingParams
from transformers import AutoTokenizer
MODEL = "Danny-Dasilva/Qwen3-14B-antidoom-W4A16-GPTQ-g32-DSpark"
DRAFT = "deepseek-ai/dspark_qwen3_14b_block7"
llm = LLM(
model=MODEL,
max_model_len=8192,
attention_backend="FLASHINFER",
speculative_config={
"method": "dspark",
"model": DRAFT,
"num_speculative_tokens": 7, # draft block size; keep at 7
"attention_backend": "TRITON_ATTN",
},
)
tok = AutoTokenizer.from_pretrained(MODEL)
prompt = tok.apply_chat_template([{"role": "user", "content": "Hello!"}],
add_generation_prompt=True, tokenize=False)
print(llm.generate([prompt], SamplingParams(temperature=0.0, max_tokens=256))[0].outputs[0].text)
Requires a vLLM build with DSpark support (merged to vLLM main 2026-07-02).
temperature=0.01, prompt sources filtered to code/instruction-following (3-8× higher loop yield
than math/QA), pair budget 58 (5h generation cap; the model is extremely loop-resistant), num_epochs=10, lr=3e-5, lora_r=128, max_seq_length=3072, load_in_4bit for the training pass (14B bf16 + optimizer exceeds 32 GB), with the built-in chosen_win ≥ 0.4 early stop. Note: vLLM needs gpu_memory_utilization≈0.70 for this target+draft pair on 32 GB.Qwen/Qwen3-14B (bf16) → antidoom FTPO LoRA merge → GPTQ W4A16 g32