| """run point-1. usage: python3 generate.py ["prompt"] [--temp T] [--n N] [--seed S]""" |
| import json, math, random, struct, sys |
|
|
| BYTES_VOCAB = [10] + list(range(32, 127)) |
| H = {b: max(-1.0, min(1.0, (b - 79) / 47.0)) for b in BYTES_VOCAB} |
|
|
|
|
| def load_w(path="model.safetensors"): |
| with open(path, "rb") as f: |
| (hlen,) = struct.unpack("<Q", f.read(8)) |
| header = json.loads(f.read(hlen)) |
| f.seek(8 + hlen) |
| (w,) = struct.unpack("<f", f.read(4)) |
| nparams = sum(math.prod(t["shape"]) for k, t in header.items() if not k.startswith("__")) |
| assert nparams == 1, nparams |
| return w |
|
|
|
|
| def main(): |
| args = sys.argv[1:] |
| prompt = next((a for a in args if not a.startswith("-")), "") |
| temp = float(args[args.index("--temp") + 1]) if "--temp" in args else 0.8 |
| count = int(args[args.index("--n") + 1]) if "--n" in args else 200 |
| seed = int(args[args.index("--seed") + 1]) if "--seed" in args else 42 |
| random.seed(seed) |
|
|
| w = load_w() |
| print(f"point-1 | parameters: 1 | w = {w:+.6f}") |
| if prompt: |
| print(f"prompt : {prompt!r}") |
| weights = [math.exp(w * H[b] / temp) for b in BYTES_VOCAB] |
| out = "".join(chr(random.choices(BYTES_VOCAB, weights=weights)[0]) for _ in range(count)) |
| print("-" * 52) |
| print(out) |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
