Update README.md

README.md

@@ -16,36 +16,35 @@ datasets:
 metrics:
 - accuracy
 pipeline_tag: text-generation
+base_model:
+- samueljayasingh/rune-0.3b-base
+- samueljayasingh/rune-0.3b-sft
 ---
 
 # Rune-R1 (351M) — GRPO Reasoning Model
 
-**Rune-R1** is a ~351M parameter decoder-only transformer trained from scratch and
-aligned for math reasoning via a three-stage pipeline:
+**Rune-R1** is a ~351M parameter decoder-only transformer trained from scratch and aligned for math reasoning via a three-stage pipeline:
 
 ```
 Pretrain (FineWeb-Edu)  →  SFT (distilled CoT format)  →  GRPO (RLVR on math correctness)
 ```
 
-This repository holds the final checkpoint: the GRPO-tuned policy, starting from
-[Rune-R1-SFT](https://huggingface.co/samueljayasingh/Rune-R1-sft) and optimized
-with Group Relative Policy Optimization against a verifiable, rule-based reward for
-math answer correctness. See [Rune-R1-Base](https://huggingface.co/samueljayasingh/Rune-R1-base)
-and [Rune-R1-SFT](https://huggingface.co/samueljayasingh/Rune-R1-sft) for the earlier
-pipeline stages.
+This repository holds the final checkpoint: the GRPO-tuned policy, starting from [Rune-R1-SFT](https://huggingface.co/samueljayasingh/Rune-R1-sft) and optimized with Group Relative Policy Optimization against a verifiable, rule-based reward for math answer correctness. See [Rune-R1-Base](https://huggingface.co/samueljayasingh/Rune-R1-base) and [Rune-R1-SFT](https://huggingface.co/samueljayasingh/Rune-R1-sft) for the earlier pipeline stages.
 
 ## Model Description
 
-- **Developed by**: samueljayasingh
-- **Model type**: Causal language model (text-only)
-- **Base model**: [Rune-R1-SFT](https://huggingface.co/samueljayasingh/Rune-R1-sft) (351M, chain-of-thought SFT on top of Rune-R1-Base)
-- **Fine-tuning method**: GRPO (Group Relative Policy Optimization) with PPO-style clipping and a KL penalty to a frozen reference (RLVR — reinforcement learning from verifiable rewards)
-- **Dataset**: `data/math_train.json` (math word problems with verifiable final answers), evaluated on a 50-example MATH-500 held-out subset
-- **Language**: English
-- **Tokenizer**: GPT-2 (`tiktoken`)
-- **License**: Apache 2.0
+| | |
+|---|---|
+| **Developed by** | samueljayasingh |
+| **Model type** | Causal language model (text-only) |
+| **Base model** | [Rune-R1-SFT](https://huggingface.co/samueljayasingh/Rune-R1-sft) (351M, chain-of-thought SFT on top of Rune-R1-Base) |
+| **Fine-tuning method** | GRPO (Group Relative Policy Optimization) with PPO-style clipping and a KL penalty to a frozen reference (RLVR — reinforcement learning from verifiable rewards) |
+| **Dataset** | `data/math_train.json` (math word problems with verifiable final answers), evaluated on a 50-example MATH-500 held-out subset |
+| **Language** | English |
+| **Tokenizer** | GPT-2 (`tiktoken`) |
+| **License** | Apache 2.0 |
 
-### Architecture details
+### Architecture Details
 
 | Parameter | Value |
 |---|---|
@@ -59,22 +58,16 @@ pipeline stages.
 
 ## Intended Uses & Limitations
 
-**Intended use:**
+### Intended Use
 - Research into RLVR / GRPO-style reasoning fine-tuning at small model scale.
 - Reference implementation for reward-verified RL post-training pipelines (pretrain → SFT → RL).
 - Studying reward hacking, KL-regularization tradeoffs, and reasoning-accuracy dynamics under a small RL step budget.
 
-**Limitations:**
-- Small model (351M) with a limited RL budget (2,000 steps) — MATH-500 accuracy
-  remains low (0–4% across evaluation checkpoints; see table below) and should not
-  be compared to production-scale reasoning models.
-- Reward signal is a rule-based correctness check (`\boxed{}` extraction + symbolic
-  grading), so the model may still learn to produce well-formatted but incorrect
-  reasoning that occasionally reward-hacks the verifier.
-- Inherits base/SFT limitations: 1024-token context, ~5B pretraining tokens, no
-  broad safety/RLHF alignment beyond the math-correctness reward.
-- Not suitable for production or user-facing deployment — this is a research
-  artifact demonstrating the training pipeline, not a competitive reasoning model.
+### Limitations
+- Small model (351M) with a limited RL budget (2,000 steps) — MATH-500 accuracy remains low (0–4% across evaluation checkpoints; see table below) and should not be compared to production-scale reasoning models.
+- Reward signal is a rule-based correctness check (`\boxed{}` extraction + symbolic grading), so the model may still learn to produce well-formatted but incorrect reasoning that occasionally reward-hacks the verifier.
+- Inherits base/SFT limitations: 1024-token context, ~5B pretraining tokens, no broad safety/RLHF alignment beyond the math-correctness reward.
+- **Not suitable for production or user-facing deployment** — this is a research artifact demonstrating the training pipeline, not a competitive reasoning model.
 
 ## How to Use
 
@@ -96,12 +89,25 @@ tokens = torch.tensor([enc.encode(prompt)], dtype=torch.long)
 # See rune/generate.py in the source repo for full sampling / KV-cache generation code.
 ```
 
-The `rune` package (model definition + generation utilities) is available at the
-[Rune-R1 GitHub repository](https://github.com/samueljayasingh/Rune-R1).
+The `rune` package (model definition + generation utilities) is available at the Rune-R1 GitHub repository.
+
+## Hardware
+
+Trained end to end — pretraining, SFT, and GRPO — on a single rented GPU instance:
+
+| Component | Spec |
+|---|---|
+| GPU | 1x AMD MI300X |
+| VRAM | 192 GB |
+| vCPU | 20 |
+| RAM | 240 GB |
+| Boot disk | 720 GB NVMe SSD |
+| Scratch disk | 5 TB NVMe SSD |
+| Rate | $1.99/hr |
 
 ## Training & Evaluation
 
-### Training procedure
+### Training Procedure
 
 | Parameter | Value |
 |---|---|
@@ -117,9 +123,8 @@ The `rune` package (model definition + generation utilities) is available at the
 | Learning rate | 1e-6 |
 | Reward function | Rule-based: extract `\boxed{}` answer, symbolically grade vs. ground truth (1.0 / 0.0) |
 | Eval cadence | MATH-500 (50-example subset), every 100 steps |
-| Hardware | 1x AMD MI300X GPU |
 
-### Evaluation results
+### Evaluation Results
 
 | Metric | Value |
 |---|---|
@@ -129,9 +134,7 @@ The `rune` package (model definition + generation utilities) is available at the
 | Max single-step average reward | 0.75 |
 | Steps with nonzero reward | 152 / 2001 |
 
-MATH-500 accuracy fluctuated in the 0–4% range throughout training rather than
-improving monotonically, reflecting the small model size and limited RL budget
-rather than a fully converged reasoning model.
+MATH-500 accuracy fluctuated in the 0–4% range throughout training rather than improving monotonically, reflecting the small model size and limited RL budget rather than a fully converged reasoning model.
 
 ## Citation
 
@@ -149,4 +152,4 @@ rather than a fully converged reasoning model.
 
 - [HuggingFaceFW/fineweb-edu](https://huggingface.co/datasets/HuggingFaceFW/fineweb-edu) — pretraining corpus.
 - [rasbt/math_distill](https://huggingface.co/datasets/rasbt/math_distill) — distilled chain-of-thought SFT data.
-- [rasbt/LLMs-from-scratch](https://github.com/rasbt/LLMs-from-scratch) — architecture and the pretrain→SFT→GRPO reasoning-from-scratch recipe this pipeline is adapted from.
+- [rasbt/LLMs-from-scratch](https://github.com/rasbt/LLMs-from-scratch) — architecture and the pretrain → SFT → GRPO reasoning-from-scratch recipe this pipeline is adapted from.