Update README.md

added more benchmarks. added more info.

README.md

@@ -34,16 +34,20 @@ tags:
 
 # Vulcan-1-Coder (GGUF)
 
-A 2.5B coding and reasoning model built by using Integer Space Artificial Selection on **[CherryPit 1.2](https://huggingface.co/DRUPEs/CherryPit-1.2-2B)**, an already excellent model made by
-the team at DRUPEs.
+![image](https://cdn-uploads.huggingface.co/production/uploads/6aa007cf4ae94de9b1aaa30c/uzEAciENqItRjhT0bwydS.png)
 
-Vulcan is an **Integer Space** model: it was generated and tested as a `.gguf`, using a new technique called
-**Integer Space Artificial Selection**. Details of the method will be published later.
+A 2.5B coding and reasoning model built by using Integer Space Artificial Selection on **[CherryPit 1.2](https://huggingface.co/DRUPEs/CherryPit-1.2-2B)**, 
+an already excellent model made by the team at DRUPEs. This is a technique I created to solve many problems I faced when wanting to create models on a basic 
+graphics card. So we work in integer space, rather than weight space. Vulcan is an **Integer Space** model: it was generated and tested as a `.gguf`. Details
+of the method will be published later and released for all. I'm working hard to put everything together. 
 
 **Made in an afternoon.** Vulcan was produced in a single afternoon along with about 30 other models (12 in its
 class), all generated and tested in integer space, on a laptop with a **6 GB VRAM** GPU, using only my own tools.
 No training runs, no cluster.
 
+Every run, lasting just a few hours, resulted in success. The techique can work for all types of models. It's a data-led technique where the training is
+squeezed out rather than training it on more data. If the information exists in the model, we select for it. 
+
 Named after Vulcan, the Roman god of fire and the forge, the smith who crafts tools for the gods.
 
 **Safetensors version: coming soon.**
@@ -57,15 +61,22 @@ Named after Vulcan, the Roman god of fire and the forge, the smith who crafts to
 ## Results
 
 Notes: This model has not been fully tested or trained further after selection. It may have flaws or issues, for example one time it send a bunch of symbols.
-But generally it scores high.
+But generally it scores high. This is with no tools AND quantised. 
+
+### AIME 2025
+
+Run 1:
+21 / 30 = 70.0% with thinking on, up to 24K tokens, temperature 1.0, 1 sample, Q8_0. Benchmarking harness. No tools.
+Per-problem results: Will add soon to a site I'm building.
 
 ### AIME 2026
 
-**23 / 30 = 76.7%** with thinking on, up to 24K tokens, temperature 1.0, 1 sample, Q4_K_M.
-Per-problem results: [`aime2026_results.md`](aime2026_results.md).
+Run 1:
+**23 / 30 = 76.7%** with thinking on, up to 24K tokens, temperature 1.0, 1 sample, Q4_K_M. Benchmarking harness. No tools.
+Per-problem results:  Will add soon to a site I'm building.
 
 Run 2: 
-**26 / 30 = 86.7%** with thinking on, up to 24k tokens, temperature 1.0, 1 sample, Q8, selectre
+Running now, with thinking on, up to 32k tokens, temperature 1.0, 1 sample, Q8. Benchmarking harness. No tools.
 
 ### Selection league
 
@@ -99,21 +110,40 @@ llama-server -m Vulcan-1-Coder-Q4_K_M-league.gguf -ngl 99 -c 32768 --jinja
 
 For questions, contact ryan@filioai.com
 
-## Problems I've solved
-
-1) Instant generation of a brand new GGUF file (takes seconds to minutes for larger models, done during the trials so there's no downtime.)
-2) Natural selection replicated with great success, in just a few rounds we see success using a specific technique. 
-3) All done in quantised space without ever making a safetensors file. After, the GGUF can be reversed to safetensors. 
-4) Tensor analysis techniques taking into account the architecture of the model and the shape of it.
-5) Much more that is unrelated to the scope of this model, but it's what made this possible - a lot of analysis helped along with thousands of files of data about models.
-6) **This doesn't add to the model. It replaces parts of the base model. The natural selection writes over the models data, it doesn't add more to it, its genetics mutate.**
-
 ## Thanks
 
 - **The team at DRUPEs** for [CherryPit 1.2](https://huggingface.co/DRUPEs/CherryPit-1.2-2B), the base of Vulcan.
 - **NANI-Nithin** for the [OpenCodeReasoning-2 LoRA adapter](https://huggingface.co/NANI-Nithin/MiniCPM5-2B-OpenCodeReasoning2-LoRA).
-- **NVIDIA** for the [OpenCodeReasoning-2](https://huggingface.co/datasets/nvidia/OpenCodeReasoning-2) dataset that
-  adapter was trained on ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)).
+- **NVIDIA** for the [OpenCodeReasoning-2](https://huggingface.co/datasets/nvidia/OpenCodeReasoning-2) dataset that adapter was trained on ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)).
+
+## Things to come - progress
+
+#### Paper and files released - 50%
+
+Hand writing the paper and hand-making the diagrams without AI so I'll release it as soon as I can as a pre-print. 
+I will remove secrets from the folder and clean it up. Folder currently has thousands of files and I'd say about 25% are relevant. 
+
+#### More benchmarks - ?
+
+Whenever I feel up to it, adding some now. 
+
+#### Quant perplexity-solver
+
+Find the quants with the best perplexity
+
+#### Further processing -  30%
+
+I've discovered another technique that works to post-train one of these models. I'm still perfecting it but I'm already getting good results. I think the models can be optimised further.
+
+#### Qwen 27b - 20%
+
+Started the analysis process.
+
+#### STEM - 50%
+
+Static Embedding. **Experimental native runtime** for the `.stem` model format and embedding engine. The current runtime remains intentionally narrow: 
+**static embedding tables, token IDs, FP32 or INT8, mean pooling and optional L2**. It is not yet a general-purpose encoder/ONNX/GGUF runner.
+It creates a .STEM format that is faster, smaller, and more accurate.
 
 ## Licence