> ## Documentation Index > Fetch the complete documentation index at: https://docs.r3al.ai/llms.txt > Use this file to discover all available pages before exploring further. # Command reference > Every r3al command and flag. `--json` and `--quiet` are accepted by every command, before or after the subcommand name. Emit exactly one JSON object on stdout and nothing else. Progress commentary moves to stderr, so piping stdout stays safe. Suppress progress output. Errors are still reported. Print the installed `r3alai` version. *** ## Free commands No plan runs consumed. All but `usage` work offline. ### `r3al export ` Convert a model to ONNX, which is what quantization takes. Free and local — a Hub export downloads weights but uploads nothing. ```bash theme={null} r3al export google/vit-base-patch16-224 --output-dir ./vit r3al export weights.pt --input-shape 1,3,224,224 --output-dir ./exported ``` A HuggingFace repo id (`owner/name`) or a path to a local checkpoint. An existing path always wins, so a local directory named like a repo id still works. Optimum task for a Hub model, e.g. `image-classification`. Usually inferred; pass it when the error says the task could not be determined. Force the local source type: `onnx`, `pytorch`, `ultralytics`, `tensorflow`, `paddle`, `tflite`. Otherwise detected from the file. Input shape for a local checkpoint that cannot be traced without one, e.g. `1,3,224,224`. ONNX opset. Defaults to the export layer's choice (17). Reports `output_model` for a single-file export, or `onnx_files` when an architecture exports several parts (encoder/decoder). Read those rather than constructing a filename. Hub exports need Optimum, which is not a base dependency: `pip install 'r3alai[export-hf]'`. Optimum owns the per-architecture export logic, so deferring to it is what makes arbitrary Hub models work. ### `r3al login` Verify an API key and store it for later commands. ```bash theme={null} export R3AL_API_KEY=r3l_live_... && r3al login r3al login --status r3al login --logout ``` Report which key is in use and where it came from. Changes nothing. Read the key from stdin, so it never enters shell history. Remove the stored key. The key comes from `$R3AL_API_KEY` or stdin, is validated against the platform before being written, and is stored at `~/.config/r3al/credentials.json` with `0600` permissions. There is deliberately no `--key` flag. Anything in argv is visible in shell history and in `ps` output to every process on the machine. ### `r3al inspect ` Describe a local model. Nothing is uploaded. Run this first: it catches "wrong file", "not actually ONNX" and "no such path" for free. Path to a model file. ONNX gets the full graph analysis; other formats report size and type only. Returns `input_name`, `input_shape`, `opset`, `node_count`, a `top_ops` histogram, `conv_heavy` (true when over 15% of nodes are convolutions), and `recommended_method`. ```bash theme={null} $ r3al inspect model.onnx model model.onnx format onnx opset 17 input_name images input_shape [1, 3, 224, 224] node_count 7 recommended_method ptq_static top_ops Conv×2, Relu×2, GlobalAveragePool×1, Flatten×1, Gemm×1 ``` ### `r3al validate` Check a configuration locally against the same schema the platform uses. Exits `2` on an impossible combination. `ptq_static` or `ptq_dynamic`. Ignored when `--mode qat`. `minmax`, `percentile` or `entropy`. ```bash theme={null} $ r3al validate --method ptq_dynamic --bitwidth 4 error: ... 4-bit needs the activation ranges that only calibration provides ``` ### `r3al methods` List the available methods, whether each needs calibration data, and what each quantizes. Offline. ### `r3al usage` The plan and remaining runs for your API key. The only free command that needs the network, so it doubles as a key check — exits `3` when the key is missing or invalid. ### `r3al skill` Install the bundled agent skill, or print where it lives. Offline; downloads nothing, so it works without access to any repository. Which agent's skills directory to install into. Explicit destination, overriding `--agent`. Print the bundled skill's directory instead of copying it. Overwrite an existing installation. ### `r3al benchmark ` Measure both models on **this machine** with ONNX Runtime. Never uploads anything, never spends a run. The original, unquantized model. A quantized bundle directory, as produced by `quantize` or `qat`, **or** an explicit `.onnx` path. Prefer the directory: a downloaded deliverable's filename carries a job-specific prefix, so it is not predictable. Timed iterations per model. With no evaluation data this measures **speed, not accuracy** — it uses random inputs of the right shape. See [Benchmarking](/guides/benchmarking). *** ## Commands that spend a plan run Both validate everything they can locally first, then refuse to submit without `--yes`. ### `r3al quantize ` Post-training quantization on R3AL's GPUs. Downloads the deliverable bundle. A local model. ONNX goes straight up; other supported formats are exported to ONNX locally first. A directory, or a comma-separated list of files. Required for `ptq_static`. Where the bundle is downloaded. `ptq_static` (best accuracy, needs calibration) or `ptq_dynamic` (no calibration, quantizes only MatMul/Gemm). `4` quantizes weights and activations to 4 bits. It halves the file again but costs real accuracy on vision models, and because the kernels dequantize to float the win is size rather than speed. Requires `ptq_static`. `minmax`, `percentile` or `entropy`. Reach for the latter two when INT8 accuracy disappoints, or whenever using `--bitwidth 4`. Clipping percentile for `--calibration-method percentile`, **in percent** — `99.99`, not `0.9999`. Required to actually submit. Without it the command exits `6` after validating. ```bash theme={null} r3al quantize model.onnx \ --calibration-data ./calibration_images \ --output-dir ./quantized \ --method ptq_static --bitwidth 8 \ --calibration-method percentile --calibration-percentile 99.99 \ --yes ``` ### `r3al qat ` Quantization-aware training: fine-tunes the model with quantization simulated in the loop. Slower than PTQ and still one plan run. See [QAT pipeline](/guides/qat-pipeline). Training images. Also used to initialise the activation clip thresholds. Held-out samples, never trained on. Strongly recommended: without them the reported fidelity is measured on the images the run trained on, which cannot distinguish learning from memorising. `1` is a smoke test. 5–10 is a reasonable first real attempt. Images per training batch. A graph exported with a **fixed** batch axis overrides this — its baked-in size is used and the run says so. Re-export with a dynamic batch axis to train in larger batches. For the clip thresholds. The weights of quantized layers train at a tenth of this, since they arrive already converged. Weight bit width, 3–8. Activation bit width, 3–8. Required to submit. ```bash theme={null} r3al qat model.onnx \ --calibration-data ./train_images \ --validation-data ./heldout_images \ --epochs 10 --yes ``` With `--validation-data` the run reports four fidelity figures — training and validation set, before and after — plus a note on what they imply about the learning rate. See [QAT pipeline](/guides/qat-pipeline) for how to read them.