forge: regenerate the model card

README.md

@@ -1,19 +1,17 @@
 ---
 license: openmdw-1.1
 license_link: https://huggingface.co/poolside/Laguna-XS-2.1/blob/main/LICENSE.md
+thumbnail: https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/hero.png
 base_model:
 - poolside/Laguna-XS-2.1
 base_model_relation: quantized
-quantized_by: AlexAtomic
+quantized_by: AtomicChat
 pipeline_tag: text-generation
 library_name: mlx
 tags:
 - atomic-chat
 - laguna
 - poolside
-- coding
-- agent
-- moe
 - mlx
 - apple-silicon
 - quantized
@@ -22,91 +20,76 @@ tags:
 <center>
 
 <div style="display:flex; justify-content:center; align-items:center; gap:2%; max-width:560px; margin:0 auto;">
-<a href="https://atomic.chat"><img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/pill_atomic.png" alt="Atomic Chat" style="width:100%; height:auto; max-width:186px;"></a>
-<a href="https://discord.gg/8wGSsvmg4V"><img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/pill_discord_v3.png" alt="Discord" style="width:100%; height:auto; max-width:184px;"></a>
-<a href="https://github.com/AtomicBot-ai/Atomic-Chat"><img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/pill_github_v3.png" alt="GitHub" style="width:100%; height:auto; max-width:141px;"></a>
+<a href="https://atomic.chat" style="flex:0 1 auto; min-width:0;"><img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/pill_atomic_v3.png" alt="Atomic Chat" style="width:100%; height:auto; max-width:186px;"></a>
+<a href="https://discord.gg/8wGSsvmg4V" style="flex:0 1 auto; min-width:0;"><img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/pill_discord_v3.png" alt="Join Discord" style="width:100%; height:auto; max-width:184px;"></a>
+<a href="https://github.com/AtomicBot-ai/Atomic-Chat" style="flex:0 1 auto; min-width:0;"><img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/pill_github_v3.png" alt="GitHub" style="width:100%; height:auto; max-width:141px;"></a>
 </div>
 
 <br/>
 
-<img src="https://poolside.ai/assets/laguna/laguna-xs2-1-banner.svg" alt="Laguna XS 2.1" style="width:100%; max-width:100%; height:auto; margin-bottom:0.6em;"/>
+<img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/hero.png" alt="Laguna XS 2.1" style="width:100%; max-width:100%; height:auto; margin-bottom:0.6em;"/>
 
 <div style="display:flex; justify-content:center; gap:0.5em;">
 <a href="https://huggingface.co/poolside/Laguna-XS-2.1"><strong>Base model: poolside/Laguna-XS-2.1</strong></a>
 </div>
 </center>
 
-**Laguna XS 2.1**, quantized to **MLX (5-bit)** by [Atomic Chat](https://atomic.chat) for Apple Silicon. Built straight from poolside's original weights. Runs fully offline on your Mac.
+**Laguna XS 2.1**, self-quantized to MLX by [Atomic Chat](https://atomic.chat). Built straight from Poolside's original weights with a per-tensor importance matrix, so this is not a repack of somebody else's files. Runs fully offline.
 
 ## Highlights
 
-- **33B total / 3B active** Mixture-of-Experts for agentic coding and long-horizon work on a local machine.
-- **Mixed attention layout:** 40 layers, 10 global + 30 sliding-window (3:1 ratio), sigmoid gating with per-layer rotary scales.
-- **256 experts + 1 shared expert**, sliding window of 512 tokens.
-- **262,144-token context.**
-- **Native interleaved reasoning**, enable or disable per request.
-- **Upgraded from Laguna XS.2**: +5.4% on SWE-bench Multilingual and stronger terminal-style performance.
+- **33.4B parameters**: the weights this repo quantizes.
+- **Context length**: 262,144 tokens (256K), as published by Poolside.
+- **40 layers**: Mixture-of-Experts, hybrid sliding-window (512) and global attention.
+- **Full imatrix ladder**: every quant is calibrated with an importance matrix.
+- **Mixed SWA and global attention layout**: Laguna XS 2.1 uses sigmoid gating with per-layer rotary scales, enabling mixed SWA (Sliding Window Attention) and global attention layers in a 3:1 ratio (across 40 total layers).
+- **KV cache in FP8**: KV cache quantized to FP8, reducing memory per token.
+- **Native reasoning support**: Interleaved thinking between tool calls with support for enabling and disabling thinking per-request.
 
 > [!NOTE]
-> These are **MLX** builds for Apple Silicon (M-series), quantized from the original weights, not a repack. Laguna's architecture runs on **[mlx-vlm](https://github.com/Blaizzy/mlx-vlm)** (0.6.3+) as a text model; stock `mlx-lm` does not yet include it.
+> These MLXs are **self-quantized from the original weights**, not a repack. The importance matrix keeps low-bit quants closer to the full-precision model.
 
 ## Model Overview
 
 | Property | Value |
 |---|---|
 | Base model | `poolside/Laguna-XS-2.1` |
-| Total parameters | 33B (3B active per token) |
-| Architecture | Laguna MoE, mixed sliding-window/global attention |
-| Experts | 256 + 1 shared |
-| Layers | 40 (10 global, 30 sliding-window) |
+| Parameters | 33.4B |
+| Layers | 40 |
+| Experts | 256 routed (top-8) |
 | Sliding window | 512 tokens |
-| Context length | 262,144 |
-| Optimizer | Muon |
-| This repo | MLX quants (3-8 bit) for Apple Silicon, built from the original weights with mlx-vlm. |
+| Context length | 262,144 tokens (256K) |
+| Vocabulary | 100,352 |
+| Modalities | Text |
+| Architecture | Mixture-of-Experts, 256 experts (top-8), hybrid sliding-window (512) and global attention, 48 attention heads over 8 KV heads, `LagunaForCausalLM` |
+| This repo | MLX weights |
 
-<img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/benchmark.png" alt="Laguna XS 2.1 benchmarks" style="width:100%; max-width:900px;"/>
-
-Scores are poolside's published results for the full-precision base `poolside/Laguna-XS-2.1`. The MLX quants run the same model locally; lower bit-widths trade a little accuracy for size and speed.
-
-## This quant
-
-This repo is the **5-bit** MLX build (~21 GB). The full ladder (5/6/8-bit) lives in the [Laguna XS 2.1 collection](https://huggingface.co/collections/AtomicChat/laguna-xs-21-6a468ed7a585a924f9fe40e0).
+<img src="https://huggingface.co/AtomicChat/Laguna-XS-2.1-MLX-5bit/resolve/main/benchmark.png" alt="Laguna XS 2.1 benchmark scores" style="width:100%; max-width:900px;"/>
 
+Scores are Poolside's published results for the base `poolside/Laguna-XS-2.1`, not our own measurements. Quantization preserves the large majority of this; `Q4_K_M` and up stay close to full precision.
 
 ## Get started
 
-- **[Atomic Chat](https://atomic.chat):** open the app, search `AtomicChat/Laguna-XS-2.1-MLX-5bit`, pick a quant, hit **Use this model**.
-- **mlx-vlm (generate):**
-  ```bash
-  pip install -U mlx-vlm
-  python -m mlx_vlm generate --model AtomicChat/Laguna-XS-2.1-MLX-5bit-5bit \
-      --prompt "Write a Python retry wrapper with exponential backoff." \
-      --max-tokens 512 --temperature 1.0
-  ```
-- **mlx-vlm (OpenAI-compatible server):**
-  ```bash
-  python -m mlx_vlm server --model AtomicChat/Laguna-XS-2.1-MLX-5bit-5bit --host 0.0.0.0 --port 8080
-  # POST http://localhost:8080/v1/chat/completions  with  "model": "6bit"
-  ```
-
-Reasoning is native and on by default. Start the server with `--enable-thinking` (optionally `--thinking-budget N`) to keep it; omit the flag for direct, non-reasoning replies.
+- **[Atomic Chat](https://atomic.chat):** search `AtomicChat/Laguna-XS-2.1-MLX-5bit` and hit **Use this model**.
+- **mlx-lm:** `mlx_lm.generate --model AtomicChat/Laguna-XS-2.1-MLX-5bit --prompt "Hello" --max-tokens 512`
+- **Server:** `mlx_lm.server --model AtomicChat/Laguna-XS-2.1-MLX-5bit --port 8080`
 
 ## Best practices
 
 | Parameter | Value |
 |---|---|
 | temperature | 1.0 |
+| top_p | 1 |
 | top_k | 20 |
-| top_p | 1.0 |
+| min_p | 0.0 |
 
-poolside's benchmark settings. For agentic coding, keep reasoning enabled and preserve prior thinking blocks across turns.
+Poolside's recommended sampling configuration for `poolside/Laguna-XS-2.1`.
 
 ## How these were made
 
-1. Download `poolside/Laguna-XS-2.1` (original BF16 weights).
-2. Quantize each rung with `python -m mlx_vlm convert --hf-path poolside/Laguna-XS-2.1 -q --q-bits <N> --q-group-size 64`.
+1. Download `poolside/Laguna-XS-2.1` (original weights).
+2. Convert and quantize with `mlx_lm.convert` on our pipeline.
 
 ## License
 
-Released by poolside under the OpenMDW-1.1 license, which permits free use, modification and redistribution with attribution. MLX conversion by Atomic Chat. This is an unofficial community quantization and is not endorsed by poolside; the original `LICENSE.md` and notices of origin are retained in each quant folder.
-
+Original model by Poolside, released under the OpenMDW-1.1 license. Full terms: [OpenMDW-1.1](https://huggingface.co/poolside/Laguna-XS-2.1/blob/main/LICENSE.md). Quantized by Atomic Chat.