> ## Documentation Index
> Fetch the complete documentation index at: https://docs.aevyra.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Recipe

> The single artifact Forge proposes, mutates, and keeps or reverts.

A **recipe** is the complete deployment specification Forge searches
over. Every experiment produces exactly one recipe; the best one is
written to `best_recipe.yaml` at the end of a run.

## Structure

```yaml theme={null}
model: meta-llama/Llama-3.2-1B-Instruct
hardware:
  vendor: nvidia
  gpu_type: Tesla T4
  count: 1
  memory_gb_per_gpu: 15
config:                          # Layer 1 — vLLM serving args
  max_num_seqs: 256
  max_num_batched_tokens: 8192
  block_size: 16
  gpu_memory_utilization: 0.9
  enable_prefix_caching: true
  enable_chunked_prefill: true
  swap_space: 4
  kv_cache_dtype: auto
  tensor_parallel_size: 1
  pipeline_parallel_size: 1
quant:                           # Layer 2 — quantization (v0: defaults only)
  method: bf16
  kv_cache_quant: none
kernels: []                      # Layer 3 — custom kernels (v0: empty)
generation: 3
parent_id: a1b2c3d4
id: e5f6a7b8
```

## Layers

**Layer 1 — Config** (`config:`) is the primary search space in v0.
Forge tunes vLLM serving arguments that control batching, memory, and
caching behaviour. These have the highest leverage per experiment
because they require no recompilation.

**Layer 2 — Quantization** (`quant:`) is scaffolded but not yet
implemented. In v0.2+ Forge will tune INT4/FP8/INT8 methods and KV
cache precision jointly with Layer 1.

**Layer 3 — Kernel synthesis** (`kernels:`) hooks into AutoKernel for
custom op synthesis. Planned for v0.3+.

## Key VLLMConfig fields

| Field                    | vLLM default | What it does                            |
| ------------------------ | ------------ | --------------------------------------- |
| `max_num_seqs`           | 256          | Max concurrent sequences in a batch     |
| `max_num_batched_tokens` | 8192         | Max tokens processed per forward pass   |
| `enable_prefix_caching`  | false        | Cache KV state for repeated prefixes    |
| `enable_chunked_prefill` | true         | Break long prefills into chunks         |
| `gpu_memory_utilization` | 0.9          | Fraction of GPU VRAM for KV cache       |
| `kv_cache_dtype`         | auto         | KV cache precision (auto/fp8/fp16/bf16) |
| `tensor_parallel_size`   | 1            | Number of GPUs for tensor parallelism   |

## Lineage

Each recipe records its `parent_id` and `generation`. This lets Forge
detect convergence, build a diff between any two recipes, and render
a clean audit trail.

```python theme={null}
from aevyra_forge.recipe import Recipe

r1 = Recipe.from_yaml(open("best_recipe.yaml").read())
diff = r1.diff(baseline)
# {"enable_prefix_caching": {"from": False, "to": True},
#  "max_num_seqs": {"from": 256, "to": 128}}
```