{ "total": 12, "limit": 24, "offset": 0, "sort": "recency", "models": [ { "slug": "smolvlm2-256m-video-instruct", "name": "HuggingFaceTB/SmolVLM2-256M-Video-Instruct", "summary": "A 256M-parameter vision-language model that watches video, not just stills.", "task": "video-text-to-text", "format": "safetensors", "parameters": 256484928, "file": { "name": "model.safetensors", "bytes": 1025998224 }, "download_filename": "SmolVLM2-256M-Video-Instruct.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/HuggingFaceTB/SmolVLM2-256M-Video-Instruct", "path": "model.safetensors" }, "sha256": "dcc8f4c00d30d70bb3028a1bfc373e1b1972e25919f8788183cae4ae72431ce2", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/smolvlm2-256m-video-instruct/download", "page_url": "https://tinymodels.co/models/smolvlm2-256m-video-instruct", "checksum_url": "https://tinymodels.co/models/smolvlm2-256m-video-instruct/checksum", "downloads": 25, "bench": [ { "metric": "one clip of 16 frames at 224x224 + one question", "value": 2841.4, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "greedy decoding, transformers, float32, synthetic clip; one warm-up run then one timed run", "measured": true, "short": "1 clip + question · transformers" }, { "metric": "peak memory to load and run that input", "value": 3171, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "3171 MB peak" }, { "metric": "Video-MME", "value": 33.7, "unit": "%", "hardware": null, "source": "https://huggingface.co/HuggingFaceTB/SmolVLM2-256M-Video-Instruct", "note": "the 256M row of the card's evaluation table" }, { "metric": "MLVU", "value": 40.6, "unit": "%", "hardware": null, "source": "https://huggingface.co/HuggingFaceTB/SmolVLM2-256M-Video-Instruct", "note": "same table" }, { "metric": "MVBench", "value": 32.7, "unit": "%", "hardware": null, "source": "https://huggingface.co/HuggingFaceTB/SmolVLM2-256M-Video-Instruct", "note": "same table" } ], "added_at": "2026-09-19T08:08:40.365Z", "source": "seed" }, { "slug": "qwen3-0-6b", "name": "Qwen/Qwen3-0.6B", "summary": "The 0.6B Qwen3 dense model: on-device chat with a switchable thinking mode.", "task": "text-generation", "format": "safetensors", "parameters": 751632384, "file": { "name": "model.safetensors", "bytes": 1503300328 }, "download_filename": "Qwen3-0.6B.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/Qwen/Qwen3-0.6B", "path": "model.safetensors" }, "sha256": "f47f71177f32bcd101b7573ec9171e6a57f4f4d31148d38e382306f42996874b", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/qwen3-0-6b/download", "page_url": "https://tinymodels.co/models/qwen3-0-6b", "checksum_url": "https://tinymodels.co/models/qwen3-0-6b/checksum", "downloads": 20, "bench": [ { "metric": "one 64-token reply to a fixed prompt", "value": 469.6, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "greedy decoding, mlx-lm on the GPU, publisher bf16 weights; one warm-up run then one timed run", "measured": true, "short": "64-token reply · mlx-lm" }, { "metric": "peak memory to load and run that input", "value": 1697, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "1697 MB peak" }, { "metric": "MMLU-Redux, thinking mode", "value": 55.6, "unit": "%", "hardware": null, "source": "https://arxiv.org/abs/2505.09388", "note": "Qwen3 Technical Report, Qwen3-0.6B row of the thinking-mode table" }, { "metric": "MATH-500, thinking mode", "value": 77.6, "unit": "%", "hardware": null, "source": "https://arxiv.org/abs/2505.09388", "note": "same table" }, { "metric": "IFEval (strict prompt), thinking mode", "value": 59.2, "unit": "%", "hardware": null, "source": "https://arxiv.org/abs/2505.09388", "note": "same table" } ], "added_at": "2026-09-16T08:08:40.365Z", "source": "seed" }, { "slug": "florence-2-base-ft", "name": "microsoft/Florence-2-base-ft", "summary": "232M parameters covering captioning, OCR, detection and segmentation behind one prompt format.", "task": "image-text-to-text", "format": "safetensors", "parameters": 231567705, "file": { "name": "model.safetensors", "bytes": 463221266 }, "download_filename": "Florence-2-base-ft.safetensors", "licence": "mit", "upstream": { "repo": "https://huggingface.co/microsoft/Florence-2-base-ft", "path": "model.safetensors" }, "sha256": "58757d657ff44051314c8030b68e04cb1bb618ca9a4885418f111f6fb708185a", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/florence-2-base-ft/download", "page_url": "https://tinymodels.co/models/florence-2-base-ft", "checksum_url": "https://tinymodels.co/models/florence-2-base-ft/checksum", "downloads": 22, "bench": [ { "metric": "one 640x480 photo (coco-val2017-000000039769.jpg) + ''", "value": 2124.4, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "greedy decoding, transformers, float32; one warm-up run then one timed run", "measured": true, "short": "1 photo + prompt · transformers" }, { "metric": "peak memory to load and run that input", "value": 2109, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "2109 MB peak" }, { "metric": "COCO Caption Karpathy test CIDEr", "value": 140, "unit": "CIDEr", "hardware": null, "source": "https://huggingface.co/microsoft/Florence-2-base-ft", "note": "the fine-tuned-model table in the card" }, { "metric": "COCO detection val2017 mAP", "value": 41.4, "unit": "mAP", "hardware": null, "source": "https://huggingface.co/microsoft/Florence-2-base-ft", "note": "same table" }, { "metric": "VQAv2 test-dev accuracy", "value": 79.7, "unit": "%", "hardware": null, "source": "https://huggingface.co/microsoft/Florence-2-base-ft", "note": "same table" } ], "added_at": "2026-09-12T08:08:40.365Z", "source": "seed" }, { "slug": "smollm2-135m-instruct", "name": "HuggingFaceTB/SmolLM2-135M-Instruct", "summary": "A 135M-parameter instruction-tuned chat model small enough to run in a browser tab.", "task": "text-generation", "format": "safetensors", "parameters": 134515008, "file": { "name": "model.safetensors", "bytes": 269060552 }, "download_filename": "SmolLM2-135M-Instruct.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/HuggingFaceTB/SmolLM2-135M-Instruct", "path": "model.safetensors" }, "sha256": "5af571cbf074e6d21a03528d2330792e532ca608f24ac70a143f6b369968ab8c", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/smollm2-135m-instruct/download", "page_url": "https://tinymodels.co/models/smollm2-135m-instruct", "checksum_url": "https://tinymodels.co/models/smollm2-135m-instruct/checksum", "downloads": 21, "bench": [ { "metric": "one 64-token reply to a fixed prompt", "value": 110.2, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "greedy decoding, mlx-lm on the GPU, publisher bf16 weights; one warm-up run then one timed run", "measured": true, "short": "64-token reply · mlx-lm" }, { "metric": "peak memory to load and run that input", "value": 658, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "658 MB peak" }, { "metric": "HellaSwag (0-shot)", "value": 40.9, "unit": "%", "hardware": null, "source": "https://huggingface.co/HuggingFaceTB/SmolLM2-135M-Instruct", "note": "instruction-model table in the model card, scored with lighteval" }, { "metric": "MMLU (cloze, 0-shot)", "value": 29.3, "unit": "%", "hardware": null, "source": "https://huggingface.co/HuggingFaceTB/SmolLM2-135M-Instruct", "note": "same table" }, { "metric": "ARC (average, 0-shot)", "value": 37.3, "unit": "%", "hardware": null, "source": "https://huggingface.co/HuggingFaceTB/SmolLM2-135M-Instruct", "note": "same table" }, { "metric": "BBH (3-shot)", "value": 28.2, "unit": "%", "hardware": null, "source": "https://huggingface.co/HuggingFaceTB/SmolLM2-135M-Instruct", "note": "same table" } ], "added_at": "2026-09-09T08:08:40.365Z", "source": "seed" }, { "slug": "qwen2-5-0-5b-instruct-4bit-mlx", "name": "mlx-community/Qwen2.5-0.5B-Instruct-4bit", "summary": "4-bit MLX conversion of Qwen2.5-0.5B-Instruct, 278 MB, tuned for Apple silicon.", "task": "text-generation", "format": "mlx", "parameters": 494032768, "file": { "name": "model.safetensors", "bytes": 278064920 }, "download_filename": "Qwen2.5-0.5B-Instruct-4bit.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/mlx-community/Qwen2.5-0.5B-Instruct-4bit", "path": "model.safetensors" }, "sha256": "ddffab9cbc7bf6dde941c6724841eeca8981fcfa81ca20ff8efff1396326d153", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct-4bit-mlx/download", "page_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct-4bit-mlx", "checksum_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct-4bit-mlx/checksum", "downloads": 20, "bench": [ { "metric": "one 64-token reply to a fixed prompt", "value": 241.4, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "greedy decoding, mlx-lm on the GPU, publisher bf16 weights; one warm-up run then one timed run", "measured": true, "short": "64-token reply · mlx-lm" }, { "metric": "peak memory to load and run that input", "value": 786, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "786 MB peak" } ], "added_at": "2026-09-04T08:08:40.365Z", "source": "seed" }, { "slug": "qwen2-5-0-5b-instruct-gguf-q4-k-m", "name": "Qwen/Qwen2.5-0.5B-Instruct-GGUF", "summary": "The same 0.5B instruct model as a 491 MB Q4_K_M GGUF, ready for llama.cpp and Ollama.", "task": "text-generation", "format": "gguf", "parameters": 494032768, "file": { "name": "qwen2.5-0.5b-instruct-q4_k_m.gguf", "bytes": 491400032 }, "download_filename": "qwen2.5-0.5b-instruct-q4_k_m.gguf", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF", "path": "qwen2.5-0.5b-instruct-q4_k_m.gguf" }, "sha256": "74a4da8c9fdbcd15bd1f6d01d621410d31c6fc00986f5eb687824e7b93d7a9db", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct-gguf-q4-k-m/download", "page_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct-gguf-q4-k-m", "checksum_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct-gguf-q4-k-m/checksum", "downloads": 21, "bench": [ { "metric": "one 64-token reply to a fixed prompt", "value": 255.4, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "greedy decoding, llama.cpp; one warm-up run then one timed run", "measured": true, "short": "64-token reply · llama.cpp" }, { "metric": "peak memory to load and run that input", "value": 795, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "795 MB peak" } ], "added_at": "2026-08-31T08:08:40.365Z", "source": "seed" }, { "slug": "videomae-base-finetuned-kinetics", "name": "MCG-NJU/videomae-base-finetuned-kinetics", "summary": "VideoMAE-base at 87M parameters, labelling 400 human actions from a clip.", "task": "video-classification", "format": "safetensors", "parameters": 86534800, "file": { "name": "model.safetensors", "bytes": 346161616 }, "download_filename": "videomae-base-finetuned-kinetics.safetensors", "licence": "cc-by-nc-4.0", "upstream": { "repo": "https://huggingface.co/MCG-NJU/videomae-base-finetuned-kinetics", "path": "model.safetensors" }, "sha256": "065cfde56f97671da4d196e1a3a6d3cf9304dcad92eb671e3632a6c6cdd04f73", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/videomae-base-finetuned-kinetics/download", "page_url": "https://tinymodels.co/models/videomae-base-finetuned-kinetics", "checksum_url": "https://tinymodels.co/models/videomae-base-finetuned-kinetics/checksum", "downloads": 22, "bench": [ { "metric": "one clip of 16 frames at 224x224", "value": 214.2, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "one forward pass, transformers, float32, synthetic clip; one warm-up run then one timed run", "measured": true, "short": "1 clip (16 frames) · transformers" }, { "metric": "peak memory to load and run that input", "value": 976, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "976 MB peak" }, { "metric": "Kinetics-400 test top-1", "value": 80.9, "unit": "%", "hardware": null, "source": "https://huggingface.co/MCG-NJU/videomae-base-finetuned-kinetics", "note": "the evaluation results the model card states" }, { "metric": "Kinetics-400 test top-5", "value": 94.7, "unit": "%", "hardware": null, "source": "https://huggingface.co/MCG-NJU/videomae-base-finetuned-kinetics", "note": "same record" } ], "added_at": "2026-08-27T08:08:40.365Z", "source": "seed" }, { "slug": "detr-resnet-50", "name": "facebook/detr-resnet-50", "summary": "DETR with a ResNet-50 backbone: 42M parameters of end-to-end object detection.", "task": "object-detection", "format": "safetensors", "parameters": 41631008, "file": { "name": "model.safetensors", "bytes": 166587896 }, "download_filename": "detr-resnet-50.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/facebook/detr-resnet-50", "path": "model.safetensors" }, "sha256": "830f5e2eeaada8c8c8281779dcc8ab12833972eb8514ed0a35be6c1d4420ad81", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/detr-resnet-50/download", "page_url": "https://tinymodels.co/models/detr-resnet-50", "checksum_url": "https://tinymodels.co/models/detr-resnet-50/checksum", "downloads": 14, "bench": [ { "metric": "one 640x480 photo (coco-val2017-000000039769.jpg)", "value": 251.1, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "one forward pass, transformers, float32; one warm-up run then one timed run", "measured": true, "short": "1 photo (640x480) · transformers" }, { "metric": "peak memory to load and run that input", "value": 1373, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "1373 MB peak" }, { "metric": "COCO 2017 validation AP", "value": 42, "unit": "AP", "hardware": null, "source": "https://huggingface.co/facebook/detr-resnet-50", "note": "the figure the model card states for this checkpoint" } ], "added_at": "2026-08-23T08:08:40.365Z", "source": "seed" }, { "slug": "qwen2-5-0-5b-instruct", "name": "Qwen/Qwen2.5-0.5B-Instruct", "summary": "Half a billion parameters of general chat and instruction following, under 1 GB in bf16.", "task": "text-generation", "format": "safetensors", "parameters": 494032768, "file": { "name": "model.safetensors", "bytes": 988097824 }, "download_filename": "Qwen2.5-0.5B-Instruct.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct", "path": "model.safetensors" }, "sha256": "fdf756fa7fcbe7404d5c60e26bff1a0c8b8aa1f72ced49e7dd0210fe288fb7fe", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct/download", "page_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct", "checksum_url": "https://tinymodels.co/models/qwen2-5-0-5b-instruct/checksum", "downloads": 16, "bench": [ { "metric": "one 64-token reply to a fixed prompt", "value": 266.8, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "greedy decoding, mlx-lm on the GPU, publisher bf16 weights; one warm-up run then one timed run", "measured": true, "short": "64-token reply · mlx-lm" }, { "metric": "peak memory to load and run that input", "value": 1451, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "1451 MB peak" }, { "metric": "Generation throughput, transformers", "value": 47.4, "unit": "tok/s", "hardware": "NVIDIA A100 80GB", "source": "https://qwen.readthedocs.io/en/v2.5/benchmark/speed_benchmark.html", "note": "BF16, batch size 1, 2048 tokens generated from a 1-token prompt" }, { "metric": "Generation throughput, vLLM", "value": 311.55, "unit": "tok/s", "hardware": "NVIDIA A100 80GB", "source": "https://qwen.readthedocs.io/en/v2.5/benchmark/speed_benchmark.html", "note": "BF16, same conditions" }, { "metric": "GPU memory, transformers BF16", "value": 0.97, "unit": "GB", "hardware": "NVIDIA A100 80GB", "source": "https://qwen.readthedocs.io/en/v2.5/benchmark/speed_benchmark.html", "note": "at a 1-token input length" } ], "added_at": "2026-08-18T08:08:40.365Z", "source": "seed" }, { "slug": "resnet-50", "name": "microsoft/resnet-50", "summary": "ResNet-50: 25M parameters, 1000 ImageNet classes, the baseline that still refuses to die.", "task": "image-classification", "format": "safetensors", "parameters": 25610152, "file": { "name": "model.safetensors", "bytes": 102482854 }, "download_filename": "resnet-50.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/microsoft/resnet-50", "path": "model.safetensors" }, "sha256": "9c6061af1f450bb0847e529fd742aa5066017be379c71bbf5546b198e5b13a1e", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/resnet-50/download", "page_url": "https://tinymodels.co/models/resnet-50", "checksum_url": "https://tinymodels.co/models/resnet-50/checksum", "downloads": 15, "bench": [ { "metric": "one 640x480 photo (coco-val2017-000000039769.jpg)", "value": 21.7, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "one forward pass, transformers, float32; one warm-up run then one timed run", "measured": true, "short": "1 photo (640x480) · transformers" }, { "metric": "peak memory to load and run that input", "value": 505, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "505 MB peak" }, { "metric": "ImageNet-1k top-1, 224px", "value": 80.38, "unit": "%", "hardware": null, "source": "https://github.com/huggingface/pytorch-image-models/blob/main/results/results-imagenet.csv", "note": "timm resnet50.a1_in1k, the ImageNet-1k weights this checkpoint was converted from" }, { "metric": "ImageNet-1k top-5, 224px", "value": 94.6, "unit": "%", "hardware": null, "source": "https://github.com/huggingface/pytorch-image-models/blob/main/results/results-imagenet.csv", "note": "same results file" } ], "added_at": "2026-08-12T08:08:40.365Z", "source": "seed" }, { "slug": "all-minilm-l6-v2", "name": "sentence-transformers/all-MiniLM-L6-v2", "summary": "The 22M-parameter sentence embedder that quietly powers a large share of local search.", "task": "sentence-similarity", "format": "safetensors", "parameters": 22713728, "file": { "name": "model.safetensors", "bytes": 90868376 }, "download_filename": "all-MiniLM-L6-v2.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2", "path": "model.safetensors" }, "sha256": "53aa51172d142c89d9012cce15ae4d6cc0ca6895895114379cacb4fab128d9db", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/all-minilm-l6-v2/download", "page_url": "https://tinymodels.co/models/all-minilm-l6-v2", "checksum_url": "https://tinymodels.co/models/all-minilm-l6-v2/checksum", "downloads": 14, "bench": [ { "metric": "one fixed 24-word sentence", "value": 5.3, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "one sentence embedded (one forward pass), transformers, float32; one warm-up run then one timed run", "measured": true, "short": "1 sentence · transformers" }, { "metric": "peak memory to load and run that input", "value": 395, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "395 MB peak" }, { "metric": "Encoding speed", "value": 14200, "unit": "sentences/s", "hardware": "NVIDIA V100 GPU", "source": "https://www.sbert.net/docs/sentence_transformer/pretrained_models.html", "note": "the sentence-transformers model table's own definition of its speed column" } ], "added_at": "2026-08-05T08:08:40.365Z", "source": "seed" }, { "slug": "distilbert-sst-2-english", "name": "distilbert/distilbert-base-uncased-finetuned-sst-2-english", "summary": "A 67M-parameter sentiment classifier: positive or negative, distilled from BERT.", "task": "text-classification", "format": "safetensors", "parameters": 66955010, "file": { "name": "model.safetensors", "bytes": 267832558 }, "download_filename": "distilbert-base-uncased-finetuned-sst-2-english.safetensors", "licence": "apache-2.0", "upstream": { "repo": "https://huggingface.co/distilbert/distilbert-base-uncased-finetuned-sst-2-english", "path": "model.safetensors" }, "sha256": "7c3919835e442510166d267fe7cbe847e0c51cd26d9ba07b89a57b952b49b8aa", "sha256_source": "upstream", "mirrored_locally": false, "download_url": "https://tinymodels.co/models/distilbert-sst-2-english/download", "page_url": "https://tinymodels.co/models/distilbert-sst-2-english", "checksum_url": "https://tinymodels.co/models/distilbert-sst-2-english/checksum", "downloads": 18, "bench": [ { "metric": "one fixed 24-word sentence", "value": 13.1, "unit": "ms", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "one sentence classified (one forward pass), transformers, float32; one warm-up run then one timed run", "measured": true, "short": "1 sentence · transformers" }, { "metric": "peak memory to load and run that input", "value": 525, "unit": "MB", "hardware": "Apple M4 Pro, 48 GB unified memory, macOS 15.6", "source": "https://github.com/mrsage-AI/tinymodels/blob/main/scripts/measure.py", "note": "peak resident set of the measuring process, runtime included", "measured": true, "short": "525 MB peak" }, { "metric": "SST-2 validation accuracy", "value": 91.06, "unit": "%", "hardware": null, "source": "https://huggingface.co/distilbert/distilbert-base-uncased-finetuned-sst-2-english", "note": "the verified evaluation recorded in the model card" }, { "metric": "SST-2 validation F1", "value": 91.37, "unit": "%", "hardware": null, "source": "https://huggingface.co/distilbert/distilbert-base-uncased-finetuned-sst-2-english", "note": "same record" }, { "metric": "SST-2 validation AUC", "value": 97.17, "unit": "%", "hardware": null, "source": "https://huggingface.co/distilbert/distilbert-base-uncased-finetuned-sst-2-english", "note": "same record" } ], "added_at": "2026-07-22T08:08:40.365Z", "source": "seed" } ] }