전체 검색
AI 모델 85021
- v6-babylm-25k-sp0.001-deu-ell-sequential↗
nikitastheo
- audio_to_audio_schrodinger_bridge↗
NVIDIA
- Riva-Translate-4B-Instruct↗
NVIDIA
- CodeBooga-34B-v0.1↗
oobabooga
It was created with the BlockMerge Gradient script, the same one that was used to create MythoMax-L2-13b, and with the same settings. The following YAML was used: I made a quick experiment where I asked a set of 3 Python and 3 Javascript questions (real world, difficult questions with nuance) to the following models: 1) This one 2) A second variant generated with `model_path1` and `model_path2` swapped in the YAML above, which I called CodeBooga-Reversed-34B-v0.1 3) WizardCoder-Python-34B-V1.0 4) Phind-CodeLlama-34B-v2
- distilbert-onnx-banking77↗
philschmid
- PentaCoder-9B↗
pragmaticcs
- Ael-Pro-40B↗
Prannesshkva
- OneLLM-Doey-V1-Llama-3.2-3B-GGUF↗
QuantFactory
- Ming-Image_GGUFs↗
realrebelai
These are not simple one-pass quants. The diffusion-model files use a **mixed-precision HQ ladder** designed to keep the most sensitive Ming-Image tensors at higher precision while aggressively quantizing the larger bulk weights. 1. Convert the original BF16 diffusion weights into a **shape-preserving BF16 GGUF master**. 2. Quantize that BF16 GGUF directly with `llama-quantize` to create donor files: - `Q8_0` - `Q6_K` - `Q5_K_M` - `Q4_K_M` - `Q3_K_M` - `Q2_K` 3. Build the final `-HQ` models by selecting tensors from those already-quantized donor files. 4. The HQ assembly step copies the existing packed GGUF tensor data directly. It does **not** dequantize and requantize each selected tensor. The `general.architecture` transport metadata used during GGUF conversion is for quantizer/loader compatibility only. It should not be interpreted as Ming-Image being a Lumina2 model.
- Meta-Llama-3.1-70B-Instruct-quantized.w4a16↗
RedHatAI
This model is a quantized version of Meta-Llama-3.1-70B-Instruct. It was evaluated on a several tasks to assess the its quality in comparison to the unquatized model, including multiple-choice, math reasoning, and open-ended text generation. Meta-Llama-3.1-70B-Instruct-quantized.w4a16 achieves 100.0% recovery for the Arena-Hard evaluation, 99.4% for OpenLLM v1 (using Meta's prompting when available), 97.4% for OpenLLM v2, 101.0% for HumanEval pass@1, and 99.2% for HumanEval+ pass@1. This model was obtained by quantizing the weights of Meta-Llama-3.1-70B-Instruct to INT4 data type. This optimization reduces the number of bits per parameter from 16 to 4, reducing the disk size and GPU memory requirements by approximately 75%. Only the weights of the linear operators within transformers blocks are quantized. Symmetric per-group quantization is applied, in which a linear scaling per group of 128 parameters maps the INT4 and floating point representations of the quantized weights. The GPTQ algorithm is applied for quantization, as implemented in the AutoGPTQ library. GPTQ used a 1% damping factor and 512 sequences of 8,192 random tokens.
