Skip to content

Instantly share code, notes, and snippets.

@fehiepsi
Created November 14, 2024 10:44
Show Gist options
  • Save fehiepsi/b7def6a77bf9ca150cf2f17f2ba1a2b5 to your computer and use it in GitHub Desktop.
Save fehiepsi/b7def6a77bf9ca150cf2f17f2ba1a2b5 to your computer and use it in GitHub Desktop.
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"from functools import partial\n",
"import matplotlib.pyplot as plt\n",
"import numpy as np\n",
"import jax\n",
"import jax.numpy as jnp\n",
"import jax.random as random\n",
"import flax\n",
"import flax.linen as nn\n",
"import optax\n",
"import numpyro\n",
"import numpyro.distributions as dist\n",
"from numpyro.ops.indexing import Vindex\n",
"import coix"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"numpyro.set_platform(\"cpu\")\n",
"coix.set_backend(\"coix.numpyro\")"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
"class LSTM_MDN(nn.Module):\n",
" n_mixture_components: int\n",
" n_features: int\n",
"\n",
" @nn.compact\n",
" def __call__(self, z_prev, x_curr, carry):\n",
" x = jnp.stack([z_prev, x_curr], axis=-1)\n",
" lstm_cell = nn.LSTMCell(name=\"lstm_cell\", features=self.n_features)\n",
" carry, x = lstm_cell(carry, x)\n",
" mu_t = nn.Dense(self.n_mixture_components)(x)\n",
" log_sigma_t = nn.Dense(self.n_mixture_components)(x)\n",
" pi_t = nn.Dense(self.n_mixture_components)(x)\n",
" return mu_t, jnp.exp(log_sigma_t), nn.softmax(pi_t), carry"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
"lstm_mdn = LSTM_MDN(n_mixture_components=3, n_features=50)"
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {},
"outputs": [],
"source": [
"def ssm_proposal(proposal, t, inputs):\n",
" if isinstance(t, int) and (t == 0):\n",
" z_t = numpyro.sample(\"z\", dist.Normal(0, 5))\n",
" else:\n",
" mu_t, sigma_t, pi_t, carry = jax.vmap(proposal, in_axes=(0, None, 0))(\n",
" inputs[\"zs\"][..., t - 1], inputs[\"xs\"][t], inputs[\"carry\"])\n",
" inputs[\"carry\"] = carry\n",
" z_t = numpyro.sample(\"z\", dist.MixtureSameFamily(\n",
" dist.Categorical(pi_t), dist.Normal(mu_t, sigma_t)))\n",
" return (inputs,)\n",
"\n",
"def f(z, t):\n",
" return z / 2 + 25 * z / (1 + z ** 2) + 8 * jnp.cos(1.2 * t)\n",
"\n",
"def g(z):\n",
" return z ** 2 / 20\n",
"\n",
"def ssm_target(t, inputs, simulate=False):\n",
" z_t_loc = jnp.where(t == 0, 0, f(inputs[\"zs\"][..., t - 1], t))\n",
" z_t_scale = jnp.where(t == 0, 5, jnp.sqrt(10))\n",
" z_t = numpyro.sample(\"z\", dist.Normal(z_t_loc, z_t_scale))\n",
" x_t = numpyro.sample(\"x\", dist.Normal(g(z_t), 1),\n",
" obs=None if simulate else inputs[\"xs\"][t])\n",
" inputs[\"zs\"] = inputs[\"zs\"].at[..., t].set(z_t)\n",
" inputs[\"xs\"] = inputs[\"xs\"].at[t].set(x_t)\n",
" return (inputs,)"
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {},
"outputs": [],
"source": [
"def nasmc(targets, proposals, *, num_targets, unroll=False):\n",
" def body_fun(i, q):\n",
" p, q = targets(i), coix.compose(coix.detach(proposals(i)), coix.resample(q))\n",
" return coix.propose(p, q, loss_fn=coix.loss.rws_loss, chain=True)\n",
"\n",
" q = coix.propose(\n",
" targets(0), coix.detach(proposals(0)), loss_fn=coix.loss.rws_loss, chain=True)\n",
" if unroll:\n",
" for i in range(1, num_targets):\n",
" q = body_fun(i, q)\n",
" return q\n",
" else:\n",
" return coix.fori_loop(1, num_targets, body_fun, q)"
]
},
{
"cell_type": "code",
"execution_count": 20,
"metadata": {},
"outputs": [],
"source": [
"def make_ssm(params, num_particles=10, T_max=1000):\n",
" network = coix.util.BindModule(lstm_mdn, params)\n",
" make_particle_plate = lambda: numpyro.plate(\"particle\", num_particles, dim=-1)\n",
" targets = lambda t: make_particle_plate()(partial(ssm_target, t))\n",
" proposals = lambda t: make_particle_plate()(partial(ssm_proposal, network, t))\n",
" program = nasmc(targets, proposals, num_targets=T_max, unroll=False)\n",
" return program"
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {},
"outputs": [],
"source": [
"def simulate_ssm(T_max=1000):\n",
" program = partial(ssm_target, 0, simulate=True)\n",
" for t in range(1, T_max):\n",
" program = coix.compose(partial(ssm_target, t, simulate=True), program)\n",
" return program"
]
},
{
"cell_type": "code",
"execution_count": 22,
"metadata": {},
"outputs": [],
"source": [
"def loss_fn(params, key, init_carry, num_particles=10, T_max=1000):\n",
" data_key, rng_key = random.split(key)\n",
" data = numpyro.handlers.seed(simulate_ssm(T_max=T_max), rng_seed=data_key)(\n",
" inputs={\"xs\": jnp.zeros(T_max), \"zs\": jnp.zeros(T_max)})[0]\n",
" assert data[\"xs\"].shape[0] == data[\"zs\"].shape[0] == T_max\n",
"\n",
" carry = jax.tree.map(lambda x: jnp.repeat(x[None], num_particles, axis=0), init_carry)\n",
" inputs = {\"zs\": jnp.zeros((num_particles, T_max)), \"xs\": data[\"xs\"], \"carry\": carry}\n",
" program = make_ssm(params, num_particles=num_particles, T_max=T_max)\n",
" _, _, metrics = coix.traced_evaluate(program, seed=rng_key)(inputs)\n",
" return metrics[\"loss\"], metrics\n",
"\n",
"def batch_loss_fn(params, key, init_carry, batch_size=10, num_particles=10, T_max=1000):\n",
" if batch_size == 1:\n",
" return loss_fn(params, key, init_carry, num_particles=num_particles, T_max=T_max)\n",
" loss, metrics = jax.vmap(\n",
" partial(loss_fn, num_particles=num_particles, T_max=T_max),\n",
" in_axes=(None, 0, None))(params, jax.random.split(key, batch_size), init_carry)\n",
" return loss.mean(), jax.tree.map(jnp.mean, metrics)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Compiling the first train step...\n",
"Time to compile a train step: 8.777919054031372\n",
"=====\n",
"Step 50 | ess 2.4568 | log_Z -187.3532 | log_density -17.5152 | log_weight -196.0697 | loss 385.9447 | squared_grad_norm 460799.5938\n",
"Step 100 | ess 14.6745 | log_Z -109.4901 | log_density -11.9905 | log_weight -116.3763 | loss 312.4891 | squared_grad_norm 100631.6953\n",
"Step 150 | ess 14.1414 | log_Z -107.8167 | log_density -13.8655 | log_weight -116.1320 | loss 287.2689 | squared_grad_norm 62360.5312\n",
"Step 200 | ess 7.5109 | log_Z -148.9827 | log_density -22.9591 | log_weight -165.1962 | loss 264.5852 | squared_grad_norm 34423.0000\n",
"Step 250 | ess 10.4280 | log_Z -117.4725 | log_density -16.2886 | log_weight -127.2894 | loss 251.7576 | squared_grad_norm 9126.8506\n",
"Step 300 | ess 13.8720 | log_Z -95.6815 | log_density -13.7289 | log_weight -104.8602 | loss 220.5114 | squared_grad_norm 722.7039\n",
"Step 350 | ess 14.2133 | log_Z -82.9974 | log_density -14.0265 | log_weight -91.2859 | loss 238.8261 | squared_grad_norm 823.7309\n",
"Step 400 | ess 11.1576 | log_Z -107.7981 | log_density -19.8433 | log_weight -121.2751 | loss 244.7899 | squared_grad_norm 797.5330\n",
"Step 450 | ess 20.9421 | log_Z -108.1516 | log_density -16.1086 | log_weight -117.6074 | loss 243.2255 | squared_grad_norm 785.7372\n",
"Step 500 | ess 13.3322 | log_Z -84.9444 | log_density -16.9784 | log_weight -96.3718 | loss 228.7092 | squared_grad_norm 429.4864\n",
"Step 550 | ess 11.4026 | log_Z -103.4894 | log_density -13.4355 | log_weight -111.0630 | loss 226.2614 | squared_grad_norm 275.1570\n",
"Step 600 | ess 17.1433 | log_Z -101.0487 | log_density -18.6916 | log_weight -113.2903 | loss 241.0913 | squared_grad_norm 408.4709\n",
"Step 650 | ess 15.1463 | log_Z -115.8731 | log_density -19.5854 | log_weight -128.9296 | loss 220.9007 | squared_grad_norm 333.0921\n",
"Step 700 | ess 14.2602 | log_Z -104.8662 | log_density -15.6090 | log_weight -114.3933 | loss 223.7206 | squared_grad_norm 320.8732\n",
"Step 750 | ess 11.3065 | log_Z -118.1048 | log_density -15.2429 | log_weight -127.5417 | loss 225.1173 | squared_grad_norm 580.1074\n",
"Step 800 | ess 13.7209 | log_Z -83.1169 | log_density -13.3460 | log_weight -90.7723 | loss 209.2690 | squared_grad_norm 357.6276\n",
"Step 850 | ess 10.1412 | log_Z -112.6723 | log_density -12.7744 | log_weight -119.0864 | loss 198.0578 | squared_grad_norm 178.7914\n",
"Step 900 | ess 11.0142 | log_Z -113.9407 | log_density -13.3378 | log_weight -121.3964 | loss 204.0146 | squared_grad_norm 201.4111\n",
"Step 950 | ess 14.5444 | log_Z -92.2560 | log_density -12.1194 | log_weight -98.4538 | loss 187.6152 | squared_grad_norm 1153.7579\n",
"Step 1000 | ess 17.6082 | log_Z -97.2455 | log_density -9.8393 | log_weight -101.0580 | loss 187.0929 | squared_grad_norm 572.7567\n"
]
}
],
"source": [
"num_particles = 100\n",
"num_steps = 1000\n",
"batch_size = 10\n",
"T_max = 100\n",
"init_carry = nn.LSTMCell(features=50).initialize_carry(random.key(1), (10,))\n",
"init_params = lstm_mdn.init(random.key(2), z_prev=0., x_curr=0., carry=init_carry)\n",
"lstm_mdn_params, _ = coix.util.train(\n",
" partial(\n",
" batch_loss_fn,\n",
" init_carry=init_carry,\n",
" num_particles=num_particles,\n",
" T_max=T_max,\n",
" batch_size=batch_size,\n",
" ),\n",
" init_params,\n",
" optax.adam(3e-4),\n",
" num_steps=num_steps,\n",
" jit_compile=True,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 42,
"metadata": {},
"outputs": [],
"source": [
"def eval_program(params, key, init_carry, num_particles=10, T_max=1000):\n",
" data_key, rng_key = random.split(key)\n",
" data = numpyro.handlers.seed(simulate_ssm(T_max=T_max), rng_seed=data_key)(\n",
" inputs={\"xs\": jnp.zeros(T_max), \"zs\": jnp.zeros(T_max)})[0]\n",
" assert data[\"xs\"].shape[0] == data[\"zs\"].shape[0] == T_max\n",
"\n",
" carry = jax.tree.map(lambda x: jnp.repeat(x[None], num_particles, axis=0), init_carry)\n",
" inputs = {\"zs\": jnp.zeros((num_particles, T_max)), \"xs\": data[\"xs\"], \"carry\": carry}\n",
" program = make_ssm(params, num_particles=num_particles, T_max=T_max)\n",
" out, _, metrics = coix.traced_evaluate(program, seed=rng_key)(inputs)\n",
" idx = metrics['log_weight'].argmax()\n",
" return data[\"zs\"], out[0][\"zs\"][idx], metrics"
]
},
{
"cell_type": "code",
"execution_count": 43,
"metadata": {},
"outputs": [],
"source": [
"true_zs, pred_zs, metrics = eval_program(\n",
" lstm_mdn_params, random.key(3), init_carry, num_particles=100, T_max=1000)"
]
},
{
"cell_type": "code",
"execution_count": 44,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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",
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plt.plot(pred_zs[-100:], color='r')\n",
"plt.plot(true_zs[-100:]);"
]
},
{
"cell_type": "code",
"execution_count": 45,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"Array(11.857068, dtype=float32)"
]
},
"execution_count": 45,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"jnp.sqrt(((pred_zs - true_zs) ** 2).mean())"
]
},
{
"cell_type": "code",
"execution_count": 46,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"Array(73.36286, dtype=float32)"
]
},
"execution_count": 46,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"metrics[\"ess\"]"
]
},
{
"cell_type": "code",
"execution_count": 47,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"Array(-1862.2279, dtype=float32)"
]
},
"execution_count": 47,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"metrics[\"log_Z\"]"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.12.6"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment