Update runners.py

runners.py

@@ -17,21 +17,15 @@ import bisect
 import functools
 import logging
 import math
-import re
+import numpy as np
+import jax
+import jax.numpy as jnp
+import haiku as hk
+import sentencepiece
 from dataclasses import dataclass
 from typing import Any, Callable, NamedTuple, Optional, Tuple
-
+from jax.experimental import pjit
-import haiku as hk
-import jax
-import jax.experimental.pjit as pjit
-import jax.numpy as jnp
-import numpy as np
-import sentencepiece
-from jax.experimental import mesh_utils
 from jax.sharding import PartitionSpec as P
-from jax.typing import ArrayLike
-
-import checkpoint as xai_checkpoint
 from model import (
     LanguageModelConfig,
     LanguageModelOutput,
@@ -40,49 +34,43 @@ from model import (
     Memory,
     KVMemory,
 )
+import checkpoint as xai_checkpoint
 
 logger = logging.getLogger(__name__)
+logger.setLevel(logging.INFO)
 rank_logger = logging.getLogger("rank")
+rank_logger.setLevel(logging.INFO)
 
 TOP_K = 8
 
-
 class SampleSettings(NamedTuple):
-    temperature: ArrayLike
+    temperature: jax.Array
-    nucleus_p: ArrayLike
+    nucleus_p: jax.Array
-    mask: ArrayLike
+    mask: jax.Array
-    # Whether a given batch element is actively used. [B]
+    active: jax.Array
-    active: ArrayLike
-
 
 class SampleOutput(NamedTuple):
-    token_id: ArrayLike
+    token_id: jax.Array
-    prob: ArrayLike
+    prob: jax.Array
-    top_k_token_ids: ArrayLike
+    top_k_token_ids: jax.Array
-    top_k_probs: ArrayLike
+    top_k_probs: jax.Array
 
-
+def insert_slice(memory: Memory, slice: Memory, length: int, i: int) -> Memory:
-def insert_slice(memory: Memory, slice, length, i):
     slice = Memory(
         layers=[
             KVMemory(layer.k, layer.v, step=jnp.array([length]))
             for layer in slice.layers
         ],
     )
-
     return jax.tree_map(lambda m, u: jax.lax.dynamic_update_index_in_dim(m, u[0], i, axis=0),
                         memory, slice)
 
-
+def pad_to_size(x: jnp.ndarray, size: int) -> jnp.ndarray:
-def pad_to_size(x, size):
     if x.shape[0] > size:
-        # Left truncate if the context is too long.
         x = x[-size:]
     return np.pad(x, [0, size - x.shape[0]], mode="constant", constant_values=0)
 
-
 def top_p_filter(logits: jax.Array, top_p: jax.Array) -> jax.Array:
-    """Performs nucleus filtering on logits."""
     assert logits.ndim == top_p.ndim, f"Expected {logits.ndim} equal {top_p.ndim}"
     sorted_logits = jax.lax.sort(logits, is_stable=False)
     sorted_probs = jax.nn.softmax(sorted_logits)
@@ -92,36 +80,27 @@ def top_p_filter(logits: jax.Array, top_p: jax.Array) -> jax.Array:
     )
     assert threshold_largest_logits.shape == logits.shape[:-1] + (1,)
     mask = logits >= threshold_largest_logits
-    # Set unused logits to -inf.
     logits = jnp.where(mask, logits, -1e10)
     return logits
 
-
 def sample_token(
     rngs: jax.random.PRNGKey,
     lm_outputs: LanguageModelOutput,
     settings: SampleSettings,
 ) -> SampleOutput:
-    # Expand the settings shape to match the logit shape.
     settings = SampleSettings(
-        temperature=jnp.expand_dims(settings.temperature, (1, 2)),  # Input [B], output [B, 1, 1].
+        temperature=jnp.expand_dims(settings.temperature, (1, 2)),
-        nucleus_p=jnp.expand_dims(settings.nucleus_p, (1, 2)),  # Input [B], output [B, 1, 1].
+        nucleus_p=jnp.expand_dims(settings.nucleus_p, (1, 2)),
-        mask=jnp.expand_dims(settings.mask, 1),  # Input [B, V], output [B, 1, V].
+        mask=jnp.expand_dims(settings.mask, 1),
-        active=settings.active,  # [B].
+        active=settings.active,
     )
     logits = lm_outputs.logits / settings.temperature.astype(lm_outputs.logits.dtype)
-    # Mask out all disallowed tokens by assigning them a near-zero probability.
     logits = jnp.where(settings.mask, logits, -1e10)
-    # Mask out all tokens that don't fall into the p-th percentile.
     logits = top_p_filter(logits, settings.nucleus_p.astype(logits.dtype))
-
     new_token = jax.vmap(jax.random.categorical)(rngs, logits)
-
     probabilities = jax.nn.softmax(logits)
     token_prob = jnp.take_along_axis(probabilities, jnp.expand_dims(new_token, 1), axis=2)
     token_prob = jnp.squeeze(token_prob, 1)
-
-    # Gather the top-k tokens and probabilities.
     top_k_probs, top_k_token_ids = jax.lax.top_k(probabilities, TOP_K)
     top_k_probs = jnp.squeeze(top_k_probs, 1)
     top_k_token_ids = jnp.squeeze(top_k_token_ids, 1)
@@ -132,19 +111,14 @@ def sample_token(
         top_k_probs,
     )
 
-
 @dataclass
 class ModelRunner:
     model: LanguageModelConfig
-
     bs_per_device: float = 2.0
-
     load_rename_rules: Optional[list[tuple[str, str]]] = None
     load_exclude_rules: Optional[list[str]] = None
-
+    rng_seed: int = 42
-    rng_seed: int = 42  # Initial rng seed.
     transform_forward: bool = False
-
     checkpoint_path: str = ""
 
     def make_forward_fn(self, mesh: Any):
@@ -159,20 +133,15 @@ class ModelRunner:
     def initialize(
         self,
         init_data,
-        local_mesh_config: tuple[int, int],
+        local_mesh_config: Tuple[int, int],
-        between_hosts_config: tuple[int, int],
+        between_hosts_config: Tuple[int, int],
     ):
         num_replicas = math.prod(between_hosts_config)
         self.model.initialize()
         self.model.fprop_dtype = jnp.bfloat16
         num_local_gpus = len(jax.local_devices())
-
-        # Calculate the global batch size from the local batch size.
         self.batch_size = int(self.bs_per_device * num_local_gpus * num_replicas)
-
-        # Calculate the batch size per host from the global batch size.
         self.local_batch_size = self.batch_size // jax.process_count()
-
         self.local_mesh_config = local_mesh_config
         self.between_hosts_config = between_hosts_config
         rank_logger.info(
@@ -181,7 +150,6 @@ class ModelRunner:
         self.mesh = make_mesh(self.local_mesh_config, self.between_hosts_config)
         self.forward = self.make_forward_fn(mesh=self.mesh)
         self.logits_fn = hk.transform(lambda tokens: self.forward(tokens)[0])
-
         self.eval_forward = self.make_forward_fn(mesh=self.mesh)
         self.logits_eval_fn = hk.transform(lambda tokens: self.eval_forward(tokens)[0])
 
@@ -234,7 +202,6 @@ class ModelRunner:
                     assert self.transform_forward
                     state_shapes = jax.eval_shape(self.init_fn, rng, init_data)
             init_state = None
-
             state = xai_checkpoint.restore(
                 checkpoint_path=self.checkpoint_path,
                 state_shapes=state_shapes,
@@ -244,11 +211,9 @@ class ModelRunner:
                 init_state=init_state,
                 params_only=True,
             )
-
             del init_state
         return state
 
-
 @dataclass
 class Request:
     prompt: str
@@ -257,18 +222,17 @@ class Request:
     rng_seed: int
     max_len: int
 
-
 @dataclass
 class InferenceRunner:
     name: str
-    runner: Any
+    runner: ModelRunner
     load: str
     tokenizer_path: str = "/tmp/xai_data/tokenizer.model"
     local_mesh_config: Tuple[int, int] = (1, 1)
     between_hosts_config: Tuple[int, int] = (1, 1)
-    pad_sizes: tuple[int] = (1024,)
+    pad_sizes: Tuple[int] = (1024,)
 
-    def get_pad_bucket(self, size):
+    def get_pad_bucket(self, size: int) -> int:
         i = bisect.bisect_left(self.pad_sizes, size)
         return self.pad_sizes[min(i, len(self.pad_sizes) - 1)]
 
@@ -276,330 +240,70 @@ class InferenceRunner:
         runner = self.runner
         self.runner.transform_forward = True
         dummy_data = dict(
-            inputs=np.zeros((1, 256), dtype=np.int32),
+            inputs=np.zeros((1, self.get_pad_bucket(512)), dtype=np.int32),
-            targets=np.zeros((1, 256), dtype=np.int32),
         )
-        runner.initialize(
+        state = runner.load_or_init(
             dummy_data,
-            local_mesh_config=self.local_mesh_config,
+            from_checkpoint=False,
-            between_hosts_config=self.between_hosts_config,
         )
-
+        runner.params = state.params
         self.tokenizer = sentencepiece.SentencePieceProcessor(model_file=self.tokenizer_path)
+        def text_to_token_ids(text):
+            ids = self.tokenizer.encode(text, out_type=int)
+            return ids
 
-        max_len = runner.model.sequence_len
+        self.text_to_token_ids = text_to_token_ids
 
-        self.vocab_size = self.runner.model.vocab_size
+    def predict(self, request: Request) -> str:
+        rng = jax.random.PRNGKey(request.rng_seed)
+        token_ids = self.text_to_token_ids(request.prompt)
+        rng, gen_rng = jax.random.split(rng)
 
-        params = runner.load_or_init(dummy_data)
+        inputs = np.array(token_ids, dtype=np.int32)[np.newaxis, :]
-        self.params = params
 
-        def pad_to_max_len(x):
+        token_ids = jnp.array(inputs)
-            if len(x.shape) > 1:
+        state = self.runner.params
-                pad_width = max_len - x.shape[1]
-                return jnp.pad(x, [(0, 0), (0, pad_width), (0, 0), (0, 0)])
-            else:
-                return x
 
-        @functools.lru_cache
-        def lm():
-            return runner.model.make(mesh=runner.mesh)
-
-        def hk_forward(
-            tokens,
-            memory=None,
-            length=None,
-            active=None,
-        ) -> LanguageModelOutput:
-            if memory is not None:
-                assert active is not None
-                layers = []
-                for l in memory.layers:
-                    # Reset steps to 0 for inactive requests to avoid unnecessary computations.
-                    step = jnp.where(active, l.step, jnp.zeros_like(l.step))
-                    layers.append(l._replace(step=step))
-                memory = memory._replace(layers=layers)
-            return lm()(tokens, memory, length=length)
-
-        def hk_sample_step(rngs, last_output: SampleOutput, memory, settings):
-            rngs, rngs_ = jax.vmap(jax.random.split, out_axes=1)(rngs)
-            lm_outputs = hk_forward(last_output.token_id, memory=memory, active=settings.active)
-            sample_result = sample_token(rngs_, lm_outputs, settings)
-            return rngs, sample_result, lm_outputs.model_state
-
-        def hk_new_memory(batch_size, sequence_len):
-            return lm().init_memory(batch_size, sequence_len)
-
-        def hk_prefill_memory(
-            rngs,
-            memory,
-            settings,
-            last_output,
-            prompt,
-            length,
-            rng_seed,
-            new_settings,
-            i,
-        ):
-            rng = jax.random.PRNGKey(seed=rng_seed)
-            rng, rng_ = jax.random.split(rng)
-
-            # Allocate new memory for this sample. The memory length is equal to the length of the
-            # prompt.
-            slice = hk_new_memory(1, prompt.shape[0])
-
-            # Move the settings for this individual batch entry into the joint settings tensor.
-            settings = jax.tree_map(
-                lambda o, v: jax.lax.dynamic_update_index_in_dim(o, v, i, axis=0),
-                settings,
-                new_settings,
-            )
-
-            # Get the settings for the batch entry from the joint settings tensor.
-            settings_slice = jax.tree_map(lambda t: jnp.expand_dims(t[i], axis=0), settings)
-
-            # Process the first n-1 tokens of the prompt.
-            lm_outputs = hk_forward(
-                jnp.expand_dims(prompt, 0),
-                memory=slice,
-                length=jnp.expand_dims(length, 0),
-                active=settings_slice.active,
-            )
-
-            # The forward pass doesn't correctly set the `step` counter inside the memory. Manually
-            # override it so `hk_forward` uses the correct context length in the next call.
-            slice = lm_outputs.model_state
-            slice = slice._replace(
-                layers=[l._replace(step=jnp.array([length])) for l in slice.layers]
-            )
-
-            # Sample the actual output token.
-            rng_ = jnp.expand_dims(rng_, 0)
-            new_output = sample_token(rng_, lm_outputs, settings_slice)
-
-            # Update the KV cache/memory.
-            slice = jax.tree_map(pad_to_max_len, slice)
-            memory = insert_slice(memory, slice, length, i)
-
-            rng = jnp.expand_dims(rng, 0)
-            rngs = jax.lax.dynamic_update_index_in_dim(rngs, rng, i, axis=0)
-
-            # Move the network outputs for this batch entry into the joint output tensor.
-            last_output = jax.tree_util.tree_map(
-                lambda last, new: jax.lax.dynamic_update_index_in_dim(last, new, i, axis=0),
-                last_output,
-                new_output,
-            )
-            return rngs, last_output, memory, settings
-
-        sample_step_ = hk.without_apply_rng(hk.transform(hk_sample_step))
-        prefill_memory_ = hk.without_apply_rng(hk.transform(hk_prefill_memory))
-        new_memory_ = hk.without_apply_rng(hk.transform(hk_new_memory))
-        forward_ = hk.without_apply_rng(hk.transform(hk_forward))
-
-        rng = jax.random.PRNGKey(42)
-        dummy_tokens = jnp.zeros((1, max_len), jnp.int32)
-
-        with runner.mesh:
-            shapes = jax.eval_shape(forward_.init, rng, dummy_tokens)
-
-        self.params_sharding = jax.tree_util.tree_map_with_path(
-            apply_rules(runner.model.partition_rules()),
-            shapes,
-        )
-
-        ds = P("data")
-        ms = runner.model.model.get_memory_sharding()
-        self.sample_step = pjit.pjit(
-            sample_step_.apply,
-            in_shardings=(self.params_sharding, None, ds, ms, None),
-            out_shardings=(None, ds, ms),
-            donate_argnums=3,
-        )
-        self.prefill_memory = pjit.pjit(
-            functools.partial(prefill_memory_.apply),
-            in_shardings=(
-                self.params_sharding,
-                None,
-                ms,
-                None,
-                ds,
-                None,
-                None,
-                None,
-                None,
-                None,
-            ),
-            out_shardings=(None, ds, ms, None),
-            donate_argnums=(2,),
-        )
-        self.new_memory = pjit.pjit(
-            new_memory_.apply,
-            static_argnums=(1, 2),
-            out_shardings=ms,
-        )
-
-    def run(self):
-        """Generator that accepts prompts."""
-        runner = self.runner
-        mesh = runner.mesh
-        max_len = runner.model.sequence_len
-        batch_size = runner.batch_size
-        params = self.params
-        rngs = jax.random.split(jax.random.PRNGKey(1), batch_size)
-        with mesh:
-            memory = self.new_memory(params, batch_size, max_len)
         settings = SampleSettings(
-                temperature=np.zeros((batch_size,), dtype=np.float32),
+            temperature=jnp.array([request.temperature]),
-                nucleus_p=np.zeros((batch_size,), dtype=np.float32),
+            nucleus_p=jnp.array([request.nucleus_p]),
-                mask=np.ones((batch_size, self.vocab_size), dtype=np.int32),
+            mask=jnp.ones(token_ids.shape, dtype=bool),
-                active=np.zeros((batch_size), dtype=np.int32),
+            active=jnp.ones(token_ids.shape, dtype=bool),
-            )
-            last_output = SampleOutput(
-                token_id=np.zeros((batch_size, 1), dtype=np.int32),
-                prob=np.zeros((batch_size, 1), dtype=jnp.bfloat16),
-                top_k_token_ids=np.zeros((batch_size, TOP_K), dtype=np.int32),
-                top_k_probs=np.zeros((batch_size, TOP_K), dtype=jnp.bfloat16),
         )
 
-            prompt = np.array([300, 400, 500, 600, 600, 700, 800])
+        for _ in range(request.max_len):
-
+            lm_outputs = self.runner.eval_forward(token_ids)
-            new_settings = SampleSettings(
+            sample_output = sample_token(gen_rng, lm_outputs, settings)
-                temperature=np.float32(1),
+            new_token = sample_output.token_id
-                nucleus_p=np.float32(1),
+            token_ids = jnp.concatenate([token_ids, new_token], axis=-1)
-                mask=np.ones((self.vocab_size,), dtype=np.int32),
+            if jnp.argmax(new_token) == 0:
-                active=np.zeros((), dtype=np.int32),
-            )
-            rng_seed = np.uint64(1)
-
-            for size in self.pad_sizes:
-                if size > runner.model.sequence_len:
                 break
-                logger.info("Precompile {}".format(size))
+
-                prompt_len = len(prompt)
+        return self.tokenizer.decode(token_ids.squeeze())
-                prompt = pad_to_size(prompt, size)
+
-                rngs, last_output, memory, settings = self.prefill_memory(
+def main():
-                    params,
+    runner = ModelRunner(
-                    rngs,
+        model=LanguageModelConfig(),
-                    memory,
+        checkpoint_path="path_to_checkpoint",
-                    settings,
-                    last_output,
-                    prompt,
-                    prompt_len,
-                    rng_seed,
-                    new_settings,
-                    0,
     )
-        with runner.mesh:
+    inference_runner = InferenceRunner(
-            logger.info("Compiling...")
+        name="inference",
-            rngs, last_output, memory = self.sample_step(
+        runner=runner,
-                params, rngs, last_output, memory, settings
+        load="path_to_load",
+        tokenizer_path="path_to_tokenizer_model",
+        local_mesh_config=(1, 1),
+        between_hosts_config=(1, 1),
     )
-            logger.info("Done compiling.")
+    inference_runner.initialize()
-
+    request = Request(
-        all_tokens = []
+        prompt="Sample text",
-        free_slots = list(range(batch_size))
+        temperature=0.7,
-        requests = [None] * batch_size
+        nucleus_p=0.9,
-        first_output = [None] * batch_size
-        jax.tree_map(lambda x: x.copy_to_host_async(), last_output)
-        prev_token = last_output
-        step = 0
-        total_num_tokens = 0
-        total_num_sequences = 0
-        with mesh:
-            while True:
-                while free_slots:
-                    request: Optional[Request] = yield
-                    tokens = self.tokenizer.encode(request.prompt)
-                    temperature = request.temperature
-                    nucleus_p = request.nucleus_p
-                    rng_seed = request.rng_seed
-
-                    i = free_slots.pop()
-                    prompt = np.array(tokens, dtype=np.int32)
-                    prompt_len = len(prompt)
-                    prompt = pad_to_size(prompt, self.get_pad_bucket(prompt.shape[0]))
-                    # All tokens are allowed.
-                    mask = np.ones((self.vocab_size,), dtype=np.int32)
-
-                    new_settings = SampleSettings(
-                        temperature=np.float32(temperature),
-                        nucleus_p=np.float32(nucleus_p),
-                        mask=mask,
-                        active=np.ones((), dtype=np.int32),
-                    )
-                    rng_seed = np.uint64(rng_seed)
-                    rngs, last_output, memory, settings = self.prefill_memory(
-                        params,
-                        rngs,
-                        memory,
-                        settings,
-                        last_output,
-                        prompt,
-                        prompt_len,
-                        rng_seed,
-                        new_settings,
-                        i,
-                    )
-                    jax.tree_map(lambda x: x.copy_to_host_async(), last_output)
-                    first_output[i] = last_output
-                    requests[i] = request
-                    total_num_sequences += 1
-
-                rngs, last_output, memory = self.sample_step(
-                    params, rngs, last_output, memory, settings
-                )
-                total_num_tokens += batch_size - len(free_slots)
-
-                # prev_token should already be on the host.
-                prev_token = jax.tree_map(np.array, prev_token)
-                for i in range(batch_size):
-                    if requests[i] is not None:
-                        if first_output[i] is not None:
-                            first_output_i = jax.tree_map(np.array, first_output[i])
-                            all_tokens.append(int(first_output_i.token_id[i][0]))
-                            first_output[i] = None
-                            continue
-
-                        all_tokens.append(int(prev_token.token_id[i][0]))
-                        cont = len(all_tokens) < requests[i].max_len
-
-                        if not cont:
-                            output_str = self.tokenizer.decode(all_tokens)
-                            requests[i] = None
-                            free_slots.append(i)
-                            all_tokens = []
-                            settings = settings._replace(active=settings.active.at[i].set(0))
-                            yield output_str
-
-                jax.tree_map(lambda x: x.copy_to_host_async(), last_output)
-                prev_token = last_output
-                step += 1
-
-
-def make_mesh(
-    local_mesh_config: tuple[int, ...], between_hosts_config: tuple[int, ...]
-) -> jax.sharding.Mesh:
-    assert len(local_mesh_config) == 2
-    assert len(between_hosts_config) == 2
-    rank_logger.info("Detected %s devices in mesh", jax.device_count())
-    device_mesh = mesh_utils.create_hybrid_device_mesh(
-        local_mesh_config,
-        between_hosts_config,
-        devices=jax.devices(),
-        process_is_granule=True,
-    )
-    rank_logger.debug(re.sub("\n+", "\n", f"Job device mesh is:\n{device_mesh}"))
-    return jax.sharding.Mesh(device_mesh, ("data", "model"))
-
-
-def sample_from_model(server, prompt, max_len, temperature):
-    next(server)
-    inp = Request(
-        prompt=prompt,
-        temperature=temperature,
-        nucleus_p=1.0,
         rng_seed=42,
-        max_len=max_len,
+        max_len=100,
     )
-    return server.send(inp)
+    result = inference_runner.predict(request)
+    print(result)
+
+if __name__ == "__main__":
+    main()