Zamba-7B-v1 is a hybrid model between Mamba, a state-space model, and transformers. It uses a mamba backbone with a shared transformer layer every 6 blocks. Zamba was trained using next-token prediction. It uses the Mistral v0.1 tokenizer. We came to this architecture after a series of ablations at small scales. Zamba-7B-v1 was pre-trained on 1T tokens of text and code data sourced from open web-datasets. Subsequently in a second phase, Zamba was annealed on a mixture of 50B high-quality tokens. Note: the current Huggingface implementation of Zamba performs slower than our internal implementation. We are working to fix this with the Huggingface team. In order to run optimized Mamba implementations on a CUDA device, you need to install `mamba-ssm` and `causal-conv1d`:
译文排队中 · 当前显示来源原文
输入 文本→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
输入 文本→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
分类待补充
输入 待核验→ 输出 待核验
Zamba2-1.2B-instruct is obtained from Zamba2-1.2B by fine-tuning on instruction-following and chat datasets. Specifically: 1. SFT of the base Zamba2-1.2B model on ultrachat_200k and Infinity-Instruct 2. DPO of the SFT checkpoint on ultrafeedback_binarized, orca_dpo_pairs, and OpenHermesPreferences Zamba2-1.2B-Instruct is a hybrid model composed of state-space (Mamba2) and transformer blocks.
译文排队中 · 当前显示来源原文
输入 文本→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
输入 文本→ 输出 文本
Zamba2-2.7B is a hybrid model composed of state-space and transformer blocks. It broadly follows the Zamba architecture which consists of a Mamba backbone alternating with shared transformer blocks (see diagram in Model Details). Zamba2-2.7B possesses three major improvements over Zamba1: 2.) Instead of a single shared attention block, we utilize two shared attention blocks which are interleaved in an ABAB pattern throughout the network. 3.) We apply a LoRA projector to each shared MLP block, which allows the network to specialize the MLPs at each invocation of the shared layer across depth. LoRA enables us to add depth-specialization for only a minimal increase in total parameter count.
译文排队中 · 当前显示来源原文
输入 文本→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
输入 文本→ 输出 文本
Zamba2-7B-Instruct is obtained from Zamba2-7B by fine-tuning on instruction-following and chat datasets. Zamba2-7B-Instruct long-context has been extended from 4k to 16k context by adjusting the rope frequency in the attention blocks. To install dependencies necessary to run Mamba2 kernels, install `mamba-ssm` from source (due to compatibility issues with PyTorch) as well as `causal-conv1d`:
译文排队中 · 当前显示来源原文
输入 文本→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
输入 文本→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
分类待补充
输入 待核验→ 输出 待核验
*Checkpoint has been reshaped for transformers PR, special thank you to @JJJYmmm. The original checkpoint is at Zyphra/ZAYA1-8B-legacy* ZAYA1-8B is a small mixture of experts language model with 760M active parameters and 8.4B total parameters trained end-to-end by Zyphra. ZAYA1-8B sets a new standard of intelligence efficiency for its parameter count through a combination of novel architecture and innovations in pretraining and post-training. ZAYA1-8B excels at detailed long-form reasoning especially for mathematical and coding task. It punches heavily above its weight in these regimes and due to its inference efficiency and small size can be highly effective in test-time compute harnesses.
译文排队中 · 当前显示来源原文
分类待补充
输入 待核验→ 输出 待核验
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
输入 文本→ 输出 文本
ZAYA1-VL-8B is a vision-language model (VLM) built upon Zyphra's ZAYA1-8B LLM. It has state-of-the-art performance among VLMs for its size and inference efficiency. - **Paper:** ZAYA1-VL-8B Technical Report - **Code:** GitHub (zaya1-vl branch) - **Blog:** Announcement blog post ZAYA1-VL-8B builds upon and uses our ZAYA1-8B LLM as its base text decoder. We also use the Qwen2.5-VL vision encoder for the ViT. ZAYA1-VL-8B introduces two novel architectural innovations:
译文排队中 · 当前显示来源原文
输入 文本 · 图像→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
输入 文本→ 输出 音频
Zonos-v0.1 is a leading open-weight text-to-speech model trained on more than 200k hours of varied multilingual speech, delivering expressiveness and quality on par with—or even surpassing—top TTS providers. Our model enables highly natural speech generation from text prompts when given a speaker embedding or audio prefix, and can accurately perform speech cloning when given a reference clip spanning just a few seconds. The conditioning setup also allows for fine control over speaking rate, pitch variation, audio quality, and emotions such as happiness, fear, sadness, and anger. The model outputs speech natively at 44kHz. Zonos follows a straightforward architecture: text normalization and phonemization via eSpeak, followed by DAC token prediction through a transformer or hybrid backbone. An overview of the architecture can be seen below.
译文排队中 · 当前显示来源原文
输入 文本→ 输出 音频
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
输入 文本→ 输出 音频
GGUF weights for **Zyphra/ZONOS2**, ready to run with **zonos2.cpp** — a standalone ggml/GGUF C++ port of ZONOS2. The **entire pipeline** (ECAPA speaker encoder → MoE backbone → DAC vocoder) runs as native C++ linking only `libggml` + `gguf`: **no Python, no PyTorch, no CUDA-only kernels at inference time.** CPU and CUDA use the same files. ZONOS2 is Zyphra's latest text-to-speech model, trained on more than 6 million hours of varied multilingual speech, delivering expressiveness and quality on par with—or even surpassing—top TTS providers at low latency with MoE. ZONOS2 excels at high-fidelity and naturalistic voice cloning. **For model details and speech samples, check out our blog.** **A hosted version is available at cloud.zyphra.com/audio-playground.**
译文排队中 · 当前显示来源原文
输入 文本→ 输出 音频
ZR1-1.5B is a small reasoning model trained extensively on both verified coding and mathematics problems with reinforcement learning. The model outperforms Llama-3.1-70B-Instruct on hard coding tasks and improves upon the base R1-Distill-1.5B model by over 50%, while achieving strong scores on math evaluations and a 37.91% pass@1 accuracy on GPQA-Diamond with just 1.5B parameters. For training we utilized the PRIME Eurus-2-RL dataset which combines the following math and code datasets: - NuminaMath-CoT - APPS, CodeContests, TACO, and Codeforces train set We filtered math data by validating that questions are correctly graded when calling the evaluator with reference ground truth, and we removed all code examples with an empty list of test cases. Our final dataset comprised roughly 400k math + 25k code samples.
译文排队中 · 当前显示来源原文
输入 文本→ 输出 文本
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
分类待补充
输入 待核验→ 输出 待核验
ZUNA1.1 is a 380M-parameter masked diffusion autoencoder trained to reconstruct, denoise, and upsample scalp-EEG signals. Given a subset of EEG channels, ZUNA can: Compared to the original ZUNA1, ZUNA1.1 supports variable sequence-length masking (up to windows of 30 seconds). In addition to full channel reconstruction, users can now select specific segments to clean due to our new, flexible dropout training scheme. At 380M parameters, ZUNA1.1 is lightweight enough to run on a consumer GPU and can be used on CPU for many workloads. ZUNA1.1 is released under the highly permissive Apache 2.0 license.
译文排队中 · 当前显示来源原文
分类待补充
输入 待核验→ 输出 待核验
