DeepSeek-R1-Distill-Qwen-7B

CONTEXT SCALING: left null, and this is now a VERIFIED null rather than an unfilled one - rope_scaling is ABSENT from config (rope_theta 10000, sliding_window 4096, max_position_embeddings 131072). The model card advertises 131072 for this distill; where the extension comes from when the config declares no interpolation is a question the config cannot answer, and null plus this note is the honest record. Note the distinguishability from the Llama-based distills above, which DO carry llama3 rope_scaling in config: same R1 brand, same advertised 131072, different mechanism. Same Qwen2 dense architecture as the Qwen2.5-7B parent (hidden 3584, 28 layers, 28 query heads / 4 KV heads); safetensors total 7615616512. head_dim derived as 3584 / 28 = 128 (absent from config.json). gqa_ratio = 28 / 4 = 7. No rope_scaling; 131072 declared natively. Serving note: this is the most-downloaded single-GPU reasoning distill and the standard 4-bit-quantisation target; the architecture is unchanged from the base model, so quantisation kernels, KV-cache sizing and batching all transfer directly.KV bytes/token at bf16 (2 bytes/element): 28 layers x 4 KV heads x 128 head_dim x 2 (K and V) x 2 bytes = 57344 B/token = 56.0 KiB/token. At 128k context that is 7.00 GiB of KV for one sequence at bf16. SERVICEABILITY JOIN (added 2026-10-03; the architecture x engine matrix this record was missing, with a citation per cell in docs/13-model-compatibility.md): FAMILY=dense-gqa. REQUIRED KERNEL CLASS: ordinary causal attention plus GEMM/GEMV. This is the baseline every engine in this repo implements, so the join here is a floor rather than a filter: no engine record in data/engines/ states that dense GQA is unsupported. SUPPORTED BY (cited): vLLM LlamaForCausalLM / Qwen2 / Qwen3 / Mistral / Phi3ForCausalLM rows [[mdl2-vllm-supported-models-2026-10]]; TensorRT-LLM LlamaForCausalLM, Qwen2ForCausalLM, Qwen3ForCausalLM, Phi3ForCausalLM rows [[mdl2-trtllm-supported-models]]; SGLang generative table (Llama, Qwen, Mistral families) [[mdl2-sglang-generative-models]]; llama.cpp LLM_ARCH_LLAMA [[pap-arch-llamacpp-arch-registry]]. HARD CONSTRAINTS THAT DO APPLY even to dense GQA, and they are model-shape rather than architecture-class: tensor-parallel size must divide the attention head count or the engine refuses to start [[mgpu-tp-size-must-divide-attention-heads]] (affects vllm and sglang); and a GQA KV cache is the thing paged-attention machinery exists for, so an engine with no paged KV (llama.cpp's slot-based pool [[llama-cpp]], exllamav2 [[exllamav2]]) loses the concurrency lever rather than the ability to run. ONE HARDWARE-DEPENDENT KERNEL GATE THAT IS NOT ABOUT GQA: on RDNA, vLLM's custom paged-attention kernel is gated on head_size == 128 AND block_size == 16, so a dense GQA model with head_dim 256 falls to the Triton 2D fallback and decode falls off a cliff - measured 12.1 tok/s at 518 tokens of context to 4.2 tok/s at 32k, with kernel_paged_attention_2d growing 28.3x [[rdna-triton-paged-attn-decode-cliff]]. Check head_dim before blaming the card.