docs: add hardware requirements (ES/EN)
Browse files
README.md
CHANGED
|
@@ -42,6 +42,15 @@ Fine-tune **LoRA (QLoRA 4-bit)** sobre el mismo mix curado que `daimon x`, pesad
|
|
| 42 |
|
| 43 |
`eval_loss < train_loss`, y notablemente más bajo que `daimon x` (0.633 vs 0.868) → el 4B aprende mejor.
|
| 44 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
### ⬇️ Descargar (opcional — solo si querés usarlo)
|
| 46 |
```bash
|
| 47 |
huggingface-cli download lucas-mella/Daimon-R
|
|
@@ -74,6 +83,15 @@ llama-server --model Qwen3-4B-Instruct-2507-Q4_K_M.gguf \
|
|
| 74 |
|
| 75 |
`eval_loss < train_loss`, and notably lower than `daimon x` (0.633 vs 0.868) → the 4B learns better.
|
| 76 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 77 |
### ⬇️ Download (optional)
|
| 78 |
```bash
|
| 79 |
huggingface-cli download lucas-mella/Daimon-R
|
|
|
|
| 42 |
|
| 43 |
`eval_loss < train_loss`, y notablemente más bajo que `daimon x` (0.633 vs 0.868) → el 4B aprende mejor.
|
| 44 |
|
| 45 |
+
### 💻 Requisitos de hardware
|
| 46 |
+
| Recurso | Mínimo | Recomendado |
|
| 47 |
+
|---|---|---|
|
| 48 |
+
| Disco (GGUF Q4_K_M + LoRA) | ~2.5 GB | ~2.5 GB |
|
| 49 |
+
| RAM | 8 GB | 16 GB |
|
| 50 |
+
| GPU | opcional (corre en CPU, más lento) | ≥ 4 GB VRAM (ej. GTX 1650) |
|
| 51 |
+
|
| 52 |
+
Más pesado que `daimon x` por ser 4B, pero la cuantización **Q4_K_M** lo hace correr en GPUs modestas.
|
| 53 |
+
|
| 54 |
### ⬇️ Descargar (opcional — solo si querés usarlo)
|
| 55 |
```bash
|
| 56 |
huggingface-cli download lucas-mella/Daimon-R
|
|
|
|
| 83 |
|
| 84 |
`eval_loss < train_loss`, and notably lower than `daimon x` (0.633 vs 0.868) → the 4B learns better.
|
| 85 |
|
| 86 |
+
### 💻 Hardware requirements
|
| 87 |
+
| Resource | Minimum | Recommended |
|
| 88 |
+
|---|---|---|
|
| 89 |
+
| Disk (Q4_K_M GGUF + LoRA) | ~2.5 GB | ~2.5 GB |
|
| 90 |
+
| RAM | 8 GB | 16 GB |
|
| 91 |
+
| GPU | optional (CPU works, slower) | ≥ 4 GB VRAM (e.g. GTX 1650) |
|
| 92 |
+
|
| 93 |
+
Heavier than `daimon x` (it's 4B), but **Q4_K_M** quantization lets it run on modest GPUs.
|
| 94 |
+
|
| 95 |
### ⬇️ Download (optional)
|
| 96 |
```bash
|
| 97 |
huggingface-cli download lucas-mella/Daimon-R
|