Openai_gpt-oss-20b-CODER-NEO-CODE-DI-MATRIX-GGUF

DavidAU/Openai_gpt-oss-20b-CODER-NEO-CODE-DI-MATRIX-GGUF

Openai_gpt-oss-20b-CODER-NEO-CODE-DI-MATRIX-GGUF is a text generation model from DavidAU. Use it when you need the model to write or continue text. The card lists the license as apache-2.0.

<small<B style="font-family: 'Arial Black', sans-serif; font-weight: 900; color: f1c40f; text-decoration: none; text-shadow: 1px 1px 0 d35400, 2px 2px 0 d35400, 3px 3px 0 d35400, 4px 4px 0 d35400, 5px 5px 0 d35400, 6p…

Downloads · 30 days

4% of all-time downloads

All-time downloads

25.9K

Public

Repo size

59.2 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.gguf59.2 GB · 100%

At a glance

Task: Text Generation
License: apache-2.0
Access: Public
Created: Aug 6, 2025
Updated: Nov 9, 2026
SHA: 6d0b3f12

Try a prompt

Base models

openai/gpt-oss-20b

Spaces

localellm/WebGPU-openai-gpt-oss-20b-coder-neo-code-di-matrix-gguf

Model card

From the Hugging Face model README

<B style="font-family: 'Arial Black', sans-serif; font-weight: 900; color: #f1c40f; text-decoration: none; text-shadow: 1px 1px 0 #d35400, 2px 2px 0 #d35400, 3px 3px 0 #d35400, 4px 4px 0 #d35400, 5px 5px 0 #d35400, 6px 6px 7px rgba(0,0,0,0.4);">[DavidAU] <a href="https://huggingface.co/DavidAU" target="_blank" style="padding-left:5px">FOLLOW</a> | <a href="https://huggingface.co/DavidAU" target="_blank">ALL MODELS</a> | <a href="https://huggingface.co/DavidAU/collections" target="_blank">COLLECTIONS</a> | <a href="https://huggingface.co/models?sort=trending&search=DavidAU" target="_blank">TRENDING</a> | <a href="https://huggingface.co/DavidAU/support" target="_blank">FAQs</a> | <a href="https://huggingface.co/DavidAU/support" target="_blank">SUPPORT</a> | <a href="https://buymeacoffee.com/davidbelton" target="_blank">Buy Me A Coffee (😁)</a>

Specialized quants (including "MXFP4_MOE.gguf") for new OpenAI 20B MOE - Mixture of Experts Model at 80+ t/s.

<h2>Openai_gpt-oss-20b-CODER-NEO-CODE-DI-MATRIX-GGUF</h2> <img src="matrix2.webp" style="float:right; width:300px; height:300px; padding:10px;">

These are DUEL NEO Imatrix / CODE IMATRIX GGUFs, NEO dataset by DavidAU with "Code" dataset compiled by DavidAU.

NEO dataset improves overall performance.

CODER dataset is specifically for coding performance.

Example generation provided at bottom of this page.

DUEL ("DI")-> Separate Imatrix datasets ("NEO" and "CODER" - generated separately per model) are co-joined to create a new Imatrix dataset, which is then applied to the quants.

Model also passed "hard" coding test too (4 experts); no issues (IQ4_NL).

TEST: Forcing the model to create code with no dependencies and limits of coding short cuts, with multiple loops, and in real time with no blocking in a language that does not support it normally.

(example prompt, and generation below)

Due to quanting issues with this model (which result in oddball quant sizes / mixtures), only TESTED quants will be uploaded (at the moment).

Currently that means IQ4_NL, and MXFP4_MOE (a new OpenAI specific quant type) are available.

NEO/Coder datasets performance improvements will show the most in the IQ4_NL quants.

There are TWO "IQ4_NL" quants:

OpenAI-20B-NEO-CODE-DIMAT-IQ4_NL.gguf : DI Imatrix applied.
OpenAI-20B-NEO-CODE-DIMAT-2-IQ4_NL.gguf : DI Imatrix applied, including output tensor (also imatrixed), and embed tensor at IQ4_NL.

There are THREE NEO MXFP4_MOE quants:

OpenAI-20B-NEO-CODE-DIMAT-MXFP4_MOE2.gguf : Output tensor Q5_1 (DI Imatrix applied)
OpenAI-20B-NEO-CODE-DIMAT-MXFP4_MOE3.gguf : Output tensor IQ4_NL (DI Imatrix applied)
OpenAI-20B-NEO-CODE-DIMAT-MXFP4_MOE4.gguf : Output tensor IQ4_NL (DI Imatrix applied) AND Embed at IQ4_NL - this makes this quant the smallest version.

MXFP4_MOE quants vastly outperform (at the moment) all other quants, except IQ4_NL, Q5_1 and Q8_0 due to odd issues compressing OpenAI's 20B model due to odd "tensor" dimensions.

( IQ4_NL, Q5_1 and Q8_0 quants are compatible with OpenAI's tensor structure as of this writing. )

MXFP4_MOE quant was specifically designed for (what appears to be) loss less quantization relative to source model. Jury still out on this.

Suggested Settings (tested in Lmstudio, Beta Branch 0.3.21 ; 4 ):

Context: 8k min.
Temp 1.2+ for creative. Temp .6 (or so) for coding/general.
Rep pen 1.1, topk 40, topp .95, min p 0.05
Experts 4-8 depending on use case. (higher than 8 MAY lower quality)

Model Supports:

128k context
up to 24 experts
Tools use, browsing, etc

For my help docs, SETTING NUMBER OF EXPERTS, and other see below.

See more about this model here:

https://huggingface.co/openai/gpt-oss-20b

Please refer to their model card, especially to control "thinking" levels.

<H2>Help, Adjustments, Samplers, Parameters and More</H2>

CHANGE THE NUMBER OF ACTIVE EXPERTS:

See this document:

https://huggingface.co/DavidAU/How-To-Set-and-Manage-MOE-Mix-of-Experts-Model-Activation-of-Experts

Settings: CHAT / ROLEPLAY and/or SMOOTHER operation of this model:

In "KoboldCpp" or "oobabooga/text-generation-webui" or "Silly Tavern" ;

Set the "Smoothing_factor" to 1.5

: in KoboldCpp -> Settings->Samplers->Advanced-> "Smooth_F"

: in text-generation-webui -> parameters -> lower right.

: In Silly Tavern this is called: "Smoothing"

NOTE: For "text-generation-webui"

-> if using GGUFs you need to use "llama_HF" (which involves downloading some config files from the SOURCE version of this model)

Source versions (and config files) of my models are here:

https://huggingface.co/collections/DavidAU/d-au-source-files-for-gguf-exl2-awq-gptq-hqq-etc-etc-66b55cb8ba25f914cbf210be

OTHER OPTIONS:

Increase rep pen to 1.1 to 1.15 (you don't need to do this if you use "smoothing_factor")
If the interface/program you are using to run AI MODELS supports "Quadratic Sampling" ("smoothing") just make the adjustment as noted.

Highest Quality Settings / Optimal Operation Guide / Parameters and Samplers

This a "Class 1" model:

For all settings used for this model (including specifics for its "class"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) please see:

[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]

You can see all parameters used for generation, in addition to advanced parameters and samplers to get the most out of this model here:

[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]

<h2>Example Generation</h2>

Using default settings (coder), 4 experts and no system prompt.

Prompt:

Write a JavaScript for fetch api using 4 ais, all using streaming that does NOT use "promise.all", and processes all streams separately but in real time - that means all streams need to be processed at the same time and not interrupt each other.

Also include sending a payload to each of the 4 ais with content, system prompt, temp, topk, topp, minp, repeat penalty ; all of these parameters will be in an array for each of the four ais.

Record the entire response - chunks - from each AI in an array - text.

Also, send the chunks in real time to 4 separate text boxes on an HTML page so real time readout from each AI can be seen by the user.

IMPORTANT: All streams must be processed at the same time, not in sequence. IMPORTANT: Do not use any dependencies, nor import any module(s).

OUTPUT:

(note: Pretty print, color and some formatting lost on copy/paste)

Thought for 11.81 seconds We need to write JavaScript code that uses fetch API with streaming responses (likely using ReadableStream from Response.body.getReader()). We have 4 AI endpoints; each will be called with a payload containing content, system prompt, temp, topk, topp, minp, repeat penalty. These parameters are in an array for each of the four AIs. We need to process all streams concurrently, not sequentially, and not using Promise.all. So we can start fetch calls, get readers, then read from each reader in separate async functions that run concurrently (like starting them but not awaiting). We need to send chunks in real time to 4 separate text boxes on an HTML page. We cannot use dependencies or import modules; just plain JS. We need to record entire response chunks per AI into arrays. So we maintain array of strings for each AI, push each chunk as it arrives. Also we need to send the payload to each AI with content, system prompt, temp, topk, topp, minp, repeat penalty. These parameters will be in an array for each of the four AIs. So maybe we have a config array like: const aiConfigs = [ {url: '', params: {...}}, ... ]; But they said "parameters will be in an array for each of the four ais." So we can define arrays. We need to send payload as JSON body. Use fetch with method POST, headers: Content-Type: application/json, body: JSON.stringify(payload). Then handle streaming: Response.body is a ReadableStream; getReader(); read loop reading chunks (Uint8Array). Convert to string using TextDecoder. Append to array and update UI. We need to run all four concurrently but not use Promise.all. We can just start them sequentially but not await each before starting next. For example: function startAI(index) { ... } Then call for i=0..3: startAI(i); That will kick off fetch calls; inside we handle reading and UI updates. We need to ensure they run concurrently, so no blocking. Also record entire response chunks in an array per AI. We also need to send the chunks in real time to 4 separate text boxes on an HTML page. So we need to have <textarea id="ai1"> etc. Then update innerText or value as new chunk arrives. We must not use dependencies, so no external libs. Also we need to handle errors. Let's produce code snippet with explanation. We'll create a function fetchAndStream(aiIndex) that returns nothing but handles reading and UI updates. We also need to define arrays for each AI's parameters. For example: const aiParams = [ {url: '', content: 'Hello', systemPrompt: 'You are ...', temp: 0.7, topk: 50, topp: 0.95, minp: 0.01, repeatPenalty: 2}, ... ]; But they said "parameters will be in an array for each of the four ais." So we can define arrays like: const aiConfigs = [ {url: '', payload: {...}}, ... ]; We need to send content, system prompt, temp, topk, topp, minp, repeat penalty. So we can create payload object accordingly. Also we need to record entire response chunks in an array per AI. We'll have const responses = [[], [], [], []]; Then push each chunk string into responses[aiIndex]. We also need to update UI: document.getElementById('output'+(i+1)).value += chunk; or innerText. Let's produce final code with comments and explanation. Also we need to ensure no Promise.all. So just call startAI for each index. Also we need to handle fetch errors gracefully. Also we need to use async/await inside each function but not block others. We can do: async function streamFromAi(index) { ... } Then call streamFromAi(0); streamFromAi(1); etc. Since they are async functions, calling them without await will start them concurrently. Alternatively we can wrap in IIFE: (async () => {...})(); But simpler: just call them. We need to ensure that the fetch calls happen concurrently; but if we use await inside each function, it won't block others because each function runs independently. So it's fine. Also we need to record entire response chunks from each AI in an array - we can push chunk string into responses[index]. Now produce final code snippet with HTML skeleton and JS script. We also need to mention that the fetch API must support streaming; some browsers may not support reading body as stream? But modern ones do. We'll provide fallback for older browsers maybe but not necessary. Also we need to ensure no dependencies, so just plain JS. Let's craft final answer: Provide explanation then code snippet. OUTPUT: Below is a stand‑alone example that fulfils every requirement you listed: <ul> <li>4 independent AI calls</li> <li>All four streams are read concurrently (no <code class="western">Promise.all</code> or any other blocking construct)</li> <li>Each call receives its own payload (<code class="western">content</code>, <code class="western">systemPrompt</code>, <code class="western">temp</code>, <code class="western">topk</code>, <code class="western">topp</code>, <code class="western">minp</code>, <code class="western">repeatPenalty</code>) – the values are stored in an array of objects.</li> <li>Every chunk that arrives is pushed into a per‑AI array and written immediately to a dedicated <code class="western"><textarea></code> so the user can see the output live.</li> <li>No external libraries or imports – pure vanilla JavaScript.</li> </ul> <hr/> <h2 class="western" style="border-top: none; border-bottom: 1px solid #3b3b3b; border-left: none; border-right: none; padding-top: 0cm; padding-bottom: 0.05cm; padding-left: 0cm; padding-right: 0cm"> 1️⃣ HTML skeleton</h2> <pre class="western" style="border: 1px solid #3b3b3b; padding: 0.05cm"><code class="western"><!DOCTYPE html></code> <code class="western"><html lang="en"></code> <code class="western"><head></code> <code class="western"> </code><code class="western"><meta charset="utf-8"></code> <code class="western"> </code><code class="western"><title>Concurrent AI Streaming Demo</title></code> <code class="western"> </code><code class="western"><style></code> <code class="western"> </code><code class="western">textarea { width:100%; height:150px; margin-bottom:10px;}</code> <code class="western"> </code><code class="western"></style></code> <code class="western"></head></code> <code class="western"><body></code> <code class="western"> </code><code class="western"></code> <code class="western"> </code><code class="western"><textarea id="output0" placeholder="AI #1 …"></textarea></code> <code class="western"> </code><code class="western"><textarea id="output1" placeholder="AI #2 …"></textarea></code> <code class="western"> </code><code class="western"><textarea id="output2" placeholder="AI #3 …"></textarea></code> <code class="western"> </code><code class="western"><textarea id="output3" placeholder="AI #4 …"></textarea></code>

<code class="western"> </code><code class="western"></code> <code class="western"> </code><code class="western"><button id="startBtn">Start All 4 AIs</button></code>

<hr/> <h2 class="western" style="border-top: none; border-bottom: 1px solid #3b3b3b; border-left: none; border-right: none; padding-top: 0cm; padding-bottom: 0.05cm; padding-left: 0cm; padding-right: 0cm"> 2️⃣ JavaScript (<code class="western">ai-stream.js</code>)</h2> <pre class="western" style="border: 1px solid #3b3b3b; padding: 0.05cm"><code class="western">/* -------------------------------------------------------------</code> <code class="western"> </code><code class="western">* 1. Configuration – one object per AI</code> <code class="western"> </code><code class="western">* ------------------------------------------------------------- */</code> <code class="western">const aiConfigs = [</code> <code class="western"> </code><code class="western">{</code> <code class="western"> </code><code class="western">// URL of the first AI endpoint</code> <code class="western"> </code><code class="western">url: 'https://api.example.com/v1/chat',</code> <code class="western"> </code><code class="western">payload: {</code> <code class="western"> </code><code class="western">content: "Hello, what can you do?",</code> <code class="western"> </code><code class="western">systemPrompt: "You are a helpful assistant.",</code> <code class="western"> </code><code class="western">temp: 0.7,</code> <code class="western"> </code><code class="western">topk: 50,</code> <code class="western"> </code><code class="western">topp: 0.95,</code> <code class="western"> </code><code class="western">minp: 0.01,</code> <code class="western"> </code><code class="western">repeatPenalty: 2</code> <code class="western"> </code><code class="western">}</code> <code class="western"> </code><code class="western">},</code> <code class="western"> </code><code class="western">{</code> <code class="western"> </code><code class="western">url: 'https://api.example.com/v1/chat',</code> <code class="western"> </code><code class="western">payload: {</code> <code class="western"> </code><code class="western">content: "Tell me a joke.",</code> <code class="western"> </code><code class="western">systemPrompt: "You are a funny chatbot.",</code> <code class="western"> </code><code class="western">temp: 0.8,</code> <code class="western"> </code><code class="western">topk: 40,</code> <code class="western"> </code><code class="western">topp: 0.9,</code> <code class="western"> </code><code class="western">minp: 0.02,</code> <code class="western"> </code><code class="western">repeatPenalty: 1</code> <code class="western"> </code><code class="western">}</code> <code class="western"> </code><code class="western">},</code> <code class="western"> </code><code class="western">{</code> <code class="western"> </code><code class="western">url: 'https://api.example.com/v1/chat',</code> <code class="western"> </code><code class="western">payload: {</code> <code class="western"> </code><code class="western">content: "Explain relativity.",</code> <code class="western"> </code><code class="western">systemPrompt: "You are a physics teacher.",</code> <code class="western"> </code><code class="western">temp: 0.6,</code> <code class="western"> </code><code class="western">topk: 60,</code> <code class="western"> </code><code class="western">topp: 0.97,</code> <code class="western"> </code><code class="western">minp: 0.01,</code> <code class="western"> </code><code class="western">repeatPenalty: 3</code> <code class="western"> </code><code class="western">}</code> <code class="western"> </code><code class="western">},</code> <code class="western"> </code><code class="western">{</code> <code class="western"> </code><code class="western">url: 'https://api.example.com/v1/chat',</code> <code class="western"> </code><code class="western">payload: {</code> <code class="western"> </code><code class="western">content: "Give me a recipe for pancakes.",</code> <code class="western"> </code><code class="western">systemPrompt: "You are a cooking expert.",</code> <code class="western"> </code><code class="western">temp: 0.5,</code> <code class="western"> </code><code class="western">topk: 30,</code> <code class="western"> </code><code class="western">topp: 0.85,</code> <code class="western"> </code><code class="western">minp: 0.02,</code> <code class="western"> </code><code class="western">repeatPenalty: 2</code> <code class="western"> </code><code class="western">}</code> <code class="western"> </code><code class="western">}</code> <code class="western">];</code>

<code class="western">/* -------------------------------------------------------------</code> <code class="western"> </code><code class="western">* 3. Helper – read a stream and push chunks to UI & storage</code> <code class="western"> </code><code class="western">* ------------------------------------------------------------- */</code> <code class="western">async function streamFromAi(aiIndex) {</code> <code class="western"> </code><code class="western">const cfg = aiConfigs[aiIndex];</code> <code class="western"> </code><code class="western">const outputBox = document.getElementById(output${aiIndex});</code> <code class="western"> </code><code class="western">const decoder = new TextDecoder('utf-8');</code>

<code class="western"> </code><code class="western">try {</code> <code class="western"> </code><code class="western">// 3.1 Send the request</code> <code class="western"> </code><code class="western">const response = await fetch(cfg.url, {</code> <code class="western"> </code><code class="western">method: 'POST',</code> <code class="western"> </code><code class="western">headers: { 'Content-Type': 'application/json' },</code> <code class="western"> </code><code class="western">body: JSON.stringify(cfg.payload)</code> <code class="western"> </code><code class="western">});</code>

<code class="western"> </code><code class="western">if (!response.ok) {</code> <code class="western"> </code><code class="western">throw new Error(HTTP ${response.status} – ${response.statusText});</code> <code class="western"> </code><code class="western">}</code>

<code class="western"> </code><code class="western">// 3.2 Get the reader</code> <code class="western"> </code><code class="western">const reader = response.body.getReader();</code>

<code class="western"> </code><code class="western">// 3.3 Read loop – runs until stream ends</code> <code class="western"> </code><code class="western">while (true) {</code> <code class="western"> </code><code class="western">const { value, done } = await reader.read();</code> <code class="western"> </code><code class="western">if (done) break; // no more data</code>

<code class="western"> </code><code class="western">// Convert Uint8Array to string</code> <code class="western"> </code><code class="western">const chunkText = decoder.decode(value, { stream: true });</code>

<code class="western"> </code><code class="western">// Store the chunk in the array</code> <code class="western"> </code><code class="western">allResponses[aiIndex].push(chunkText);</code>

<code class="western"> </code><code class="western">// Push it live into the textarea</code> <code class="western"> </code><code class="western">outputBox.value += chunkText;</code> <code class="western"> </code><code class="western">}</code>

<code class="western"> </code><code class="western">// Finalise – flush any remaining buffered text</code> <code class="western"> </code><code class="western">outputBox.value += decoder.decode(); // empty decode to flush</code>

<code class="western"> </code><code class="western">} catch (err) {</code> <code class="western"> </code><code class="western">console.error(AI #${aiIndex+1} error:, err);</code> <code class="western"> </code><code class="western">outputBox.value += \n[Error] ${err.message}\n;</code> <code class="western"> </code><code class="western">}</code> <code class="western">}</code>

<code class="western">/* -------------------------------------------------------------</code> <code class="western"> </code><code class="western">* 4. Kick‑off – start all four streams concurrently</code> <code class="western"> </code><code class="western">* ------------------------------------------------------------- */</code> <code class="western">function startAllAIs() {</code> <code class="western"> </code><code class="western">// Reset UI and storage</code> <code class="western"> </code><code class="western">for (let i = 0; i < 4; ++i) {</code> <code class="western"> </code><code class="western">document.getElementById(output${i}).value = '';</code> <code class="western"> </code><code class="western">allResponses[i] = [];</code> <code class="western"> </code><code class="western">}</code>

<code class="western"> </code><code class="western">// Start each stream – the async function is called but not awaited,</code> <code class="western"> </code><code class="western">// so they run in parallel.</code> <code class="western"> </code><code class="western">for (let i = 0; i < 4; ++i) {</code> <code class="western"> </code><code class="western">streamFromAi(i); // <-- no await</code> <code class="western"> </code><code class="western">}</code> <code class="western">}</code>

<hr/> <h3 class="western" style="border: none; padding: 0cm">How it works</h3> <ol> <li>Configuration – <code class="western">aiConfigs</code> holds one object per AI, each with its own payload (<code class="western">content</code>, <code class="western">systemPrompt</code>, etc.).</li> <li>Storage – <code class="western">allResponses</code> is an array of four arrays; every chunk that arrives for a given AI is pushed into the corresponding sub‑array.</li> <li>Streaming – Inside <code class="western">streamFromAi()</code> we: <ul> <li>Send a POST request with the JSON payload.</li> <li>Grab the <code class="western">ReadableStream</code> from <code class="western">response.body</code>.</li> <li>Use its <code class="western">getReader()</code> and read in a loop (<code class="western">reader.read()</code>).</li> <li>Each chunk (a <code class="western">Uint8Array</code>) is decoded to text, appended to the per‑AI array and written immediately into the dedicated <code class="western"><textarea></code>.</li> <li>The loop ends when <code class="western">done</code> becomes true.</li> </ul> <li>Concurrent start – In <code class="western">startAllAIs()</code> we simply call <code class="western">streamFromAi(i)</code> for each index without awaiting them. Because each function is async, calling it starts its own fetch and read loop immediately; all four loops run in parallel, so the streams are processed simultaneously.</li> <li>No external dependencies – everything uses only standard browser APIs (<code class="western">fetch</code>, <code class="western">ReadableStream</code>, <code class="western">TextDecoder</code>, DOM methods).</li> </ol> <hr/> <h3 class="western" style="border: none; padding: 0cm">What you’ll see</h3> When you click Start All 4 AIs, each <code class="western"><textarea></code> will start filling up in real time as the corresponding AI streams its response. The full text of every AI is also kept in memory (the <code class="western">allResponses</code> arrays) so you can later inspect or export it if needed. Feel free to adapt the URLs, payloads, and UI layout to match your own environment. Happy streaming!

Task: Text Generation
License: apache-2.0
Languages: en
Created: Aug 6, 2025
Updated: Nov 9, 2026