This repository hosts two generations of model. v8 is current; the v7 transformer checkpoints are kept for reproducibility only. For each stock on each day, the next-day return relative to the universe average that day — not the absolute return. The output is a ranking; the intended use is a dollar-neutral long/short book (long the top names, short the bottom ones). This matters for reading the numbers: a score of +0.004 means "expected to beat the universe average by about 40 bp tomorrow", not "expected to rise 0.4%". A market-neutral model does not forecast market direction and should not be compared against an always-up baseline. - HistGradientBoostingRegressor × 3 seeds, averaged (400…
SAVRN Model Hub
Open-Weight Models
An open-weight model is an AI model whose trained weights are published for anyone to download. The weights are what the model learned in training. With a copy of them you can run the model on hardware you control and train it further on your own data.
Open weights are not the same as open source. Many publishers release the weights without the training data or code, and the license sets what you may do with the model. This library puts each model's full card, architecture, files, license and published evaluations on one page.
12,870Models
3,655Datasets
291Papers
7,691Publishers
12,980Sourced relationships
Updated 2026-10-24 · How the library is built
12,870 models, sorted by most downloaded.
ARIES predicts enzyme-conditioned reaction sites on substrate atoms using exact GROVER atom and directed-bond representations, ESM-C 6B residue representations, and a reaction-template representation. Source code, installation instructions, and the aries-inference command are available in the ARIES GitHub repository. Example data and precomputed embeddings are distributed separately in Every directory contains modelinfo.json, seed42.pt, seed43.pt, and seed44.pt. Keep the directory names unchanged: ARIES uses the metadata to validate architecture, ensemble seeds, and LoRA/base compatibility. The five specialist-hub adapters are the canonical random-split models. They use rank 16 and alpha 32…
Community Q8 conversion of Edge0/ARK-ASR-0.6B, prepared for a draft Hex integration. No retraining. Not affiliated with the original authors, and not a canonical handy-computer artifact. Requires an ARK-enabled runtime. Stock llama.cpp, whisper.cpp and released transcribe-cpp 0.2.4 cannot load it. Use the candidate runtime at revision 19f301609c06126e5071d1de5acdd95c714eec05. ark-asr-0.6b-Q80.gguf: 1,372,881,664 bytes. SHA-256: ceca6bacd9cab3b52892de6703562aeeb62dc8a4b14aab0b020706637dd127dc. The 0.6B name describes the decoder; the full model also contains the audio encoder and adapter. Q8 linear weights, F16 tied embedding/output matrix and convolutions, F32 norms/biases/frontend tensors.…
Quantized graphs for the browser runtime: - llmint4/ - Qwen3.5-0.8B + ASMR LoRA, merged and exported, MatMulNBits int4 - ttsint4/ - F5-TTS DiT step + vocoder, int4 MatMul, convs in session dtype llmfp16/ and ttsfp16/ hold the fp16 exports they were quantized from.
Download them in the Files & versions tab.
static quants of https://huggingface.co/BruceYg/ATPO-G-Qwen2.5-VL-7B-r0.2
AXQuant CUDA NVFP4 W4A4 mixed precision development preview. Converted from the original upstream BF16 source, without AWQ. Native E2M1 FP4 weights and inputs use per-16 E4M3FN scales and FP32 global scales. This is a retrieval embedding checkpoint, with no MTP or generative claim. Eligible attention and MLP matrices use NVFP4 W4A4. Embeddings and norms retain original BF16. Saved query: / passage: prefixes, bidirectional attention, mean pooling and L2 normalization. The native vLLM Ministral alias retains iscausal=false and every actual attention layer is verified as encoder-only. Factory conversion uses AXQuant's NumPy reference RTN encoder, independently checked against preserved BF16…
AXQuant CUDA NVFP4 W4A4 mixed precision development preview. Converted from the original upstream BF16 source, without AWQ. Native E2M1 FP4 weights and inputs use per-16 E4M3FN scales and FP32 global scales. This is a retrieval embedding checkpoint, with no MTP or generative claim. Eligible attention and MLP matrices use NVFP4 W4A4. Embeddings and norms retain original BF16. Saved query: / passage: prefixes, bidirectional attention, mean pooling and L2 normalization. The native vLLM Ministral alias retains iscausal=false and every actual attention layer is verified as encoder-only. Factory conversion uses AXQuant's NumPy reference RTN encoder, independently checked against preserved BF16…
AXQuant CUDA NVFP4 W4A4 mixed precision development preview. Converted from the original upstream BF16 source, without AWQ. Native E2M1 FP4 weights and inputs use per-16 E4M3FN scales and FP32 global scales. This is a retrieval embedding checkpoint, with no MTP or generative claim. All attention projections and the first/last two MLP blocks retain original BF16. Remaining MLP matrices use NVFP4 W4A4. Saved query instruction, last-token pooling, exactly one (151643), and L2 normalization. The chat (151645) is not appended. Factory conversion uses AXQuant's NumPy reference RTN encoder, independently checked against preserved BF16 source tensors. Source and pooling assets, calibration and…
AXQuant CUDA NVFP4 W4A4 mixed precision development preview. Converted from the original upstream BF16 source, without AWQ. Native E2M1 FP4 weights and inputs use per-16 E4M3FN scales and FP32 global scales. This is a retrieval embedding checkpoint, with no MTP or generative claim. All attention projections and the first/last two MLP blocks retain original BF16. Remaining MLP matrices use NVFP4 W4A4. Saved query instruction, last-token pooling, exactly one (151643), and L2 normalization. The chat (151645) is not appended. Factory conversion uses AXQuant's NumPy reference RTN encoder, independently checked against preserved BF16 source tensors. Source and pooling assets, calibration and…
AXQuant CUDA NVFP4 W4A4 mixed precision development preview. Converted from the original upstream BF16 source, without AWQ. Native E2M1 FP4 weights and inputs use per-16 E4M3FN scales and FP32 global scales. This is a retrieval embedding checkpoint, with no MTP or generative claim. All attention projections and the first/last two MLP blocks retain original BF16. Remaining MLP matrices use NVFP4 W4A4. Saved query instruction, last-token pooling, exactly one (151643), and L2 normalization. The chat (151645) is not appended. Factory conversion uses AXQuant's NumPy reference RTN encoder, independently checked against preserved BF16 source tensors. Source and pooling assets, calibration and…
CUDA-powered document layout detection. AX DocLayout locates layout regions in a decoded RGB page image and returns class labels, confidence scores and bounding boxes in the original raster's pixel coordinates. This is a CUDA runtime release using the same FP32 weights and ONNX graph as v1.1. The ONNX backend now explicitly enables ONNX Runtime's CUDA execution provider and TF32. Postprocessing is vectorized with equivalent raster geometry. The source package verifies the CUDA provider, effective TF32 policy and placement of major compute operations. CUDA setup failures raise an error; the loader does not retry with a CPU-only session. The architecture remains RT-DETRv2 rtdetrv2r50vdm, with…
This qwen25vl model was trained 2x faster with Unsloth
This mllama model was trained 2x faster with Unsloth
Blackfrost Pro Series is a line of specialized models packaged as professional tools and workflows. Each release is designed around a defined operational role instead of open-ended chat. 1. a domain-tuned model; 2. a deliberate operating boundary; 3. a harness or integration surface; 4. deployment guidance; 5. a validation and provenance record. This structure is intended to extend across model engineering, marketing operations, logistics, and other bounded professional workflows while preserving a recognizable Blackfrost product and documentation standard. Ablagent-35B-A3B is the Blackfrost Pro Series toolkit for Abliteration and Domain Weight Modification workflows. It helps an operator…
Ablagent-9B is a compact workflow assistant built by BlackfrostAI/SI for Abliteration and Domain Weight Modification work. Its job is to help an operator inspect evidence, maintain experimental discipline, and move a defined workflow forward. It is designed for tasks such as: - checkpoint archaeology, provenance, and source-fingerprint review; - paired-capture corpus hygiene and role/template-leakage checks; - direction-capture planning and interpretation; - edit-surface enumeration, target-set review, and MTP exclusion/restore discipline; - comparison of alpha/pass candidates using geometry, distribution drift, behavior, and HITL; - runtime validation, reasoning/tool-call checks…
"Abliterated Dolphin" is a result of my 3AM brain reading about technique called abliteration and then thinking what would happen if I tried to abliterate a model that is already relatively free, such as Dolphin. Heavily inspired by mlabonne's article on abliteration on how to redirect refusals and effectively remove, or ablate, censorship from a language model. There is really no deeper meaning to any of this than pure curiosity. This is basically a dumbed down version of the original Dolphin model I used as a base, as I have not done any DPOs to heal the damage caused by abliteration. Don't try to do anything meaningful with this model. Use the original Dolphin3.0-R1-Mistral-24B instead.…
Model Collections
Hand-picked starting points, each with the reason it exists.
Collection · 4 entries
Embedding models for retrieval
Sentence and document embedding models used to build retrieval systems. Dimension and sequence length matter more than size here, and both come from the publisher.
Collection · 4 entries
Models that fit on one accelerator
Models whose publisher-reported parameter count puts them within reach of a single accelerator at common precisions. Memory needed depends on precision and serving configuration, so treat the parameter count as the starting point, not the answer.
Collection · 6 entries
Open-weight text models worth knowing
Widely used open-weight language models, chosen because each one is a distinct family rather than a variant of the one above it. Selection, not a ranking.
Collection · 3 entries
Speech and audio models
Recognition and synthesis models, grouped so the two directions are easy to compare.
Open-Weight Models Explained
What is an open-weight model?
An AI model whose trained weights are published for anyone to download, so it can be run, tested and fine-tuned on hardware the user controls.
Is an open-weight model the same as open source?
Not always. Open weights means the trained model can be downloaded. Open source usually also means the training code and data are available and the license allows broad reuse. Many open-weight models release the weights only.
Can I use an open-weight model commercially?
It depends on the license. Apache 2.0 and MIT allow commercial use. Other licenses limit it, for example to non-commercial use or below a set number of users. Every model page here shows its license.
How much memory does an open-weight model need?
About two bytes per parameter at 16-bit precision, so a 7-billion-parameter model needs roughly 14 GB for its weights, plus memory for the context it processes. Each model page lists its parameter count and the size of its files.
Related SAVRN Research
The hub sits beside SAVRN's market data and infrastructure research: what models cost to run, and what it takes to run them.
SAVRN Index
What open models cost to run
The same open-weight model priced by every host that serves it, per million tokens.
Research Hub
Data center trackers and maps
Moratoriums, permits, power, water and capital behind the facilities that run these models.
Method
How the Model Hub is built
Sources, evidence labels, refresh behaviour, and the limits of every comparison here.