ashxhart commited on
Commit
d79f951
路
verified 路
1 Parent(s): 537c952

Improve onboarding on the TensorFold Hub profile

Browse files
Files changed (2) hide show
  1. README.md +15 -30
  2. index.html +33 -58
README.md CHANGED
@@ -4,45 +4,30 @@ sdk: static
4
  pinned: false
5
  ---
6
 
7
- <div align="center">
8
- <img src="./tensorfold-hero.png" alt="TensorFold" width="100%">
9
 
10
- # TensorFold
11
 
12
- ### Fast local LLM inference and tested model releases for Apple Silicon and NVIDIA CUDA.
13
 
14
- [tensorfold.dev](https://tensorfold.dev) 路 [Runtime and setup](https://github.com/ashhart/TensorFold) 路 [Model collections](https://huggingface.co/TensorFold/collections)
15
- </div>
16
 
17
- TensorFold is an open-source inference runtime and a practical model-release project. The runtime serves local models behind an OpenAI-compatible API, while this Hugging Face organization publishes checkpoints and supporting assets tested on real hardware.
18
 
19
- ## What you will find here
20
 
21
- - MLX quantized checkpoints for Apple Silicon
22
- - MTP and DFlash assets where the upstream model provides a compatible drafter
23
- - NVIDIA and DGX Spark recipes when a release has been tested there
24
- - Measured speed, memory use, runtime versions, and known limits
25
- - Clear credit, licences, and links to the original model authors
26
 
27
- ## Run with TensorFold
28
 
29
- ```bash
30
- curl -fsSL https://tensorfold.dev/install.sh | sh
31
- ```
32
 
33
- TensorFold supports macOS and Linux. See the [setup guide](https://github.com/ashhart/TensorFold) for current model families, runtime flags, and benchmark conditions.
34
 
35
- ## Choose a model for your Mac
36
 
37
- | Unified memory | Collection | Selection basis |
38
- | --- | --- | --- |
39
- | 64 GB | [Browse models](https://huggingface.co/collections/TensorFold/mlx-models-for-64gb-macs) | Published peak below 48 GB |
40
- | 128 GB | [Browse models](https://huggingface.co/collections/TensorFold/mlx-models-for-128gb-macs) | Published peak below 96 GB |
41
- | 256 GB | [Browse models](https://huggingface.co/collections/TensorFold/mlx-models-for-256gb-macs) | Published peak below 192 GB |
42
-
43
- These collections are starting points with nominal context headroom, not guarantees at maximum context. Each model card records the tested runtime, prompt, output length, memory evidence, and any compatibility caveats.
44
-
45
- TensorFold does not train the base models. Model design, training, evaluations, and upstream documentation remain the work of the original authors and contributors.
46
-
47
- [Follow TensorFold for new model releases, runtime updates, and fixes.](https://huggingface.co/TensorFold)
48
 
 
 
4
  pinned: false
5
  ---
6
 
7
+ <img src="https://huggingface.co/spaces/TensorFold/README/resolve/main/tensorfold-logo.png" alt="TensorFold logo" width="96">
 
8
 
9
+ # Welcome to TensorFold
10
 
11
+ TensorFold is an open-source inference engine that runs models on your own hardware and serves them through an OpenAI-compatible API. On Hugging Face, we publish model conversions for Apple silicon, along with previews for new model architectures.
12
 
13
+ [Setup guide](https://tensorfold.dev/guide) 路 [Browse all models](https://huggingface.co/TensorFold/models) 路 [Engine on GitHub](https://github.com/ashhart/TensorFold)
 
14
 
15
+ ## Find a model for your Mac
16
 
17
+ Start with the collection for your Mac鈥檚 unified memory.
18
 
19
+ - [64 GB Macs](https://huggingface.co/collections/TensorFold/mlx-models-for-64gb-macs-6a9fefda17932216ec9ab457)
20
+ - [128 GB Macs](https://huggingface.co/collections/TensorFold/mlx-models-for-128gb-macs-6a9ff0abd31bc9abbe7922d7)
21
+ - [256 GB Macs](https://huggingface.co/collections/TensorFold/mlx-models-for-256gb-macs-6a9ff0ef9fed7c5bdca15e9b)
 
 
22
 
23
+ New to MLX? Try [Nemotron 3.5 Lightning 4-bit](https://huggingface.co/TensorFold/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-MLX-4bit). Its card includes a quick start and the conditions for a measured Apple silicon run. Check the runtime and memory notes before downloading. Longer contexts and concurrent requests can use more memory.
24
 
25
+ ## What鈥檚 new
 
 
26
 
27
+ Tencent Hy4-flash with DFly2 is available as a checkpoint preview in [MLX 4-bit](https://huggingface.co/TensorFold/Hy4-flash-MLX-4bit-MTP) and [MLX 8-bit](https://huggingface.co/TensorFold/Hy4-flash-MLX-8bit-MTP). Conversion checks have passed, and TensorFold runtime support is pending.
28
 
29
+ ## Need a hand?
30
 
31
+ Use the **Community** tab on a model鈥檚 page for questions about that checkpoint. For engine issues, use [GitHub Issues](https://github.com/ashhart/TensorFold/issues). Include your hardware and memory, runtime version, and model ID so we can help.
 
 
 
 
 
 
 
 
 
 
32
 
33
+ The original model authors receive credit in the checkpoint cards, alongside links to the source models and their licences.
index.html CHANGED
@@ -1,61 +1,36 @@
1
  <!doctype html>
2
  <html lang="en">
3
- <head>
4
- <meta charset="utf-8" />
5
- <meta name="viewport" content="width=device-width, initial-scale=1" />
6
- <meta
7
- name="description"
8
- content="TensorFold provides fast local LLM inference and tested model releases for Apple Silicon and NVIDIA CUDA."
9
- />
10
- <title>TensorFold | Local LLM inference</title>
11
- </head>
12
- <body>
13
- <main>
14
- <header>
15
- <p><strong>OPEN-SOURCE LOCAL INFERENCE</strong></p>
16
- <img
17
- src="https://huggingface.co/spaces/TensorFold/README/resolve/main/tensorfold-logo.png"
18
- alt="TensorFold"
19
- width="120"
20
- />
21
- <h1>TensorFold</h1>
22
- <p><strong>Fast local LLM inference and tested model releases for Apple Silicon and NVIDIA CUDA.</strong></p>
23
- <p>
24
- <a href="https://tensorfold.dev">tensorfold.dev</a>
25
- 路
26
- <a href="https://github.com/ashhart/TensorFold">GitHub</a>
27
- </p>
28
- </header>
29
-
30
- <br />
31
-
32
- <p>
33
- TensorFold is an open-source inference runtime with an OpenAI-compatible API. This
34
- Hugging Face organization publishes compatible checkpoints and supporting assets tested
35
- on real hardware.
36
- </p>
37
-
38
- <br />
39
-
40
- <h2>What you will find here</h2>
41
- <ul>
42
- <li><strong>Apple Silicon:</strong> MLX quantized checkpoints with measured speed and memory use.</li>
43
- <li><strong>Drafted decoding:</strong> MTP and DFlash assets when the upstream model provides a compatible drafter.</li>
44
- <li><strong>NVIDIA CUDA:</strong> tested recipes for supported GPUs and DGX Spark systems.</li>
45
- <li><strong>Clear release notes:</strong> runtime versions, setup steps, known limits, licences, and upstream credit.</li>
46
- </ul>
47
-
48
- <br />
49
-
50
- <h2>Run TensorFold</h2>
51
- <p><code>curl -fsSL https://tensorfold.dev/install.sh | sh</code></p>
52
-
53
- <br />
54
-
55
- <p>
56
- TensorFold does not train the base models. Model design, training, evaluations, and
57
- upstream documentation remain the work of the original authors and contributors.
58
- </p>
59
- </main>
60
- </body>
61
  </html>
 
1
  <!doctype html>
2
  <html lang="en">
3
+ <head>
4
+ <meta charset="utf-8">
5
+ <meta name="viewport" content="width=device-width, initial-scale=1">
6
+ <meta name="description" content="Find a TensorFold checkpoint, check runtime support, and get started with local inference.">
7
+ <title>TensorFold on Hugging Face</title>
8
+ <style>
9
+ :root { color-scheme: light dark; }
10
+ .tensorfold-org-card { font: 1rem/1.6 system-ui, sans-serif; max-width: 44rem; margin: 2rem auto; padding: 0 1.25rem 2rem; }
11
+ .tensorfold-org-card h1, .tensorfold-org-card h2 { line-height: 1.25; }
12
+ .tensorfold-org-card h1 { margin-top: .5rem; }
13
+ </style>
14
+ </head>
15
+ <body>
16
+ <main class="tensorfold-org-card">
17
+ <img src="https://huggingface.co/spaces/TensorFold/README/resolve/main/tensorfold-logo.png" alt="TensorFold logo" width="96">
18
+ <h1>Welcome to TensorFold</h1>
19
+ <p>TensorFold is an open-source inference engine that runs models on your own hardware and serves them through an OpenAI-compatible API. On Hugging Face, we publish model conversions for Apple silicon, along with previews for new model architectures.</p>
20
+ <p><a href="https://tensorfold.dev/guide">Setup guide</a> 路 <a href="https://huggingface.co/TensorFold/models">Browse all models</a> 路 <a href="https://github.com/ashhart/TensorFold">Engine on GitHub</a></p>
21
+ <h2>Find a model for your Mac</h2>
22
+ <p>Start with the collection for your Mac鈥檚 unified memory.</p>
23
+ <ul>
24
+ <li><a href="https://huggingface.co/collections/TensorFold/mlx-models-for-64gb-macs-6a9fefda17932216ec9ab457">64 GB Macs</a></li>
25
+ <li><a href="https://huggingface.co/collections/TensorFold/mlx-models-for-128gb-macs-6a9ff0abd31bc9abbe7922d7">128 GB Macs</a></li>
26
+ <li><a href="https://huggingface.co/collections/TensorFold/mlx-models-for-256gb-macs-6a9ff0ef9fed7c5bdca15e9b">256 GB Macs</a></li>
27
+ </ul>
28
+ <p>New to MLX? Try <a href="https://huggingface.co/TensorFold/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-MLX-4bit">Nemotron 3.5 Lightning 4-bit</a>. Its card includes a quick start and the conditions for a measured Apple silicon run. Check the runtime and memory notes before downloading. Longer contexts and concurrent requests can use more memory.</p>
29
+ <h2>What鈥檚 new</h2>
30
+ <p>Tencent Hy4-flash with DFly2 is available as a checkpoint preview in <a href="https://huggingface.co/TensorFold/Hy4-flash-MLX-4bit-MTP">MLX 4-bit</a> and <a href="https://huggingface.co/TensorFold/Hy4-flash-MLX-8bit-MTP">MLX 8-bit</a>. Conversion checks have passed, and TensorFold runtime support is pending.</p>
31
+ <h2>Need a hand?</h2>
32
+ <p>Use the <strong>Community</strong> tab on a model鈥檚 page for questions about that checkpoint. For engine issues, use <a href="https://github.com/ashhart/TensorFold/issues">GitHub Issues</a>. Include your hardware and memory, runtime version, and model ID so we can help.</p>
33
+ <p>The original model authors receive credit in the checkpoint cards, alongside links to the source models and their licences.</p>
34
+ </main>
35
+ </body>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36
  </html>