AMajor/grok-1
1
0
Fork 0
mirror of https://github.com/xai-org/grok-1.git synced 2024-11-23 12:09:52 +03:00
Code Issues Packages Projects Releases Wiki Activity

Add exceptions to LanguageModel

Browse Source
This commit is contained in:
devindkim 2024-03-18 18:40:37 -04:00
parent d6d9447e2d
commit 1a0ba385eb
1 changed files with 18 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

41
model.py
View File

@ -1002,7 +1002,7 @@ class DenseBlock(hk.Module):
sharding=P("model", "data"), sharding=P("model", "data"),
mesh=self.mesh, mesh=self.mesh,
shard_axis=1, shard_axis=1,
)(h_w1 * h_v) )(h_w1 * h_v) # TODO: Document why this isn't sequential and whether it should be.
return h_dense return h_dense
@ -1036,13 +1036,10 @@ class DecoderLayer(hk.Module):
) -> DecoderOutput: ) -> DecoderOutput:
"""Transforms input embedding sequences to output embedding sequences.""" """Transforms input embedding sequences to output embedding sequences."""
def layer_norm(x): sharding = P(self.data_axis, None)
return hk_rms_norm(x)
if self.shard_activations: if self.shard_activations:
sharding = P(self.data_axis, None, self.model_axis) sharding = P(self.data_axis, None, self.model_axis)
else:
sharding = P(self.data_axis, None)
h = with_sharding_constraint(inputs, sharding) h = with_sharding_constraint(inputs, sharding)
attn_output = MHABlock( attn_output = MHABlock(
@ -1054,8 +1051,8 @@ class DecoderLayer(hk.Module):
data_axis=self.data_axis, data_axis=self.data_axis,
model_axis=self.model_axis, model_axis=self.model_axis,
)(layer_norm(h), mask, layer_memory) )(layer_norm(h), mask, layer_memory)
h_attn = attn_output.embeddings
h_attn = attn_output.embeddings
h_attn = layer_norm(h_attn) h_attn = layer_norm(h_attn)
h += h_attn h += h_attn
h = with_sharding_constraint(h, sharding) h = with_sharding_constraint(h, sharding)
@ -1165,15 +1162,17 @@ class LanguageModelConfig:
_initialized = False _initialized = False
def initialize(self): def initialize(self):
# We cannot specify [] as a default value (it is mutable), hence None.
model_config = self.model model_config = self.model
assert self.init_scale_override is None, (
"Overriding model initialize scale is supported only for predefined models." if self.model is None: # We cannot specify [] as a default value (it is mutable), hence None.
) raise ValueError("Model configuration is not set.")
if self.init_scale_override is not None:
raise ValueError("Overriding model initialize scale is supported only for predefined models.")
if self.model_size == 0: if self.model_size == 0:
self.model_size = model_config.emb_size self.model_size = model_config.emb_size
assert self.model is not None, "Model could not be initialized."
self._initialized = True self._initialized = True
return self return self
def make(self, *args, **kwargs): def make(self, *args, **kwargs):
@ -1194,7 +1193,7 @@ class LanguageModelConfig:
return LM_PARTITION_RULES + self.model.partition_rules() return LM_PARTITION_RULES + self.model.partition_rules()
def layer_norm(x, model): def layer_norm(x):
return hk_rms_norm(x) return hk_rms_norm(x)
@ -1213,17 +1212,12 @@ class LanguageModel(hk.Module):
tokens: jax.Array, tokens: jax.Array,
memory: Optional[Memory] = None, memory: Optional[Memory] = None,
*, *,
batch: Dict[str, jax.Array] = {},
last_hid_only: bool = False, last_hid_only: bool = False,
length: Optional[jax.Array] = None, length: Optional[jax.Array] = None,
) -> LanguageModelOutput: ) -> LanguageModelOutput:
"""Forward pass, producing a sequence of logits.""" """Forward pass, producing a sequence of logits."""
del batch # Unused.
config = self.config config = self.config
input_mask = jnp.greater(tokens, config.pad_token)
# Embed the input tokens and positions. # Embed the input tokens and positions.
in_out_embed = InOutEmbed( in_out_embed = InOutEmbed(
self.config.vocab_size, self.config.vocab_size,
@ -1235,6 +1229,7 @@ class LanguageModel(hk.Module):
input_embeddings, P("data", None, self.model.model_axis) input_embeddings, P("data", None, self.model.model_axis)
) )
input_embeddings *= config.embedding_multiplier_scale input_embeddings *= config.embedding_multiplier_scale
input_mask = jnp.not_equal(tokens, config.pad_token)
model_output = self.model( model_output = self.model(
input_embeddings, input_embeddings,
@ -1242,15 +1237,15 @@ class LanguageModel(hk.Module):
memory=memory, memory=memory,
) # [B, T, D] ) # [B, T, D]
embeddings, model_state = model_output.embeddings, model_output.memory embeddings, model_state = model_output.embeddings, model_output.memory
if embeddings.dtype != self.fprop_dtype:
raise ValueError(f"Expected forward propagation dtype {self.fprop_dtype} but got {embeddings.dtype} in embeddings.")
if self.model.shard_activations: if self.model.shard_activations:
embeddings = with_sharding_constraint( embeddings = with_sharding_constraint(embeddings, P("data", None, self.model.model_axis))
embeddings, P("data", None, self.model.model_axis)
)
else: else:
embeddings = with_sharding_constraint(embeddings, P("data", None)) embeddings = with_sharding_constraint(embeddings, P("data", None))
rank_logger.debug(f"Final embedding shape: {embeddings.shape}") rank_logger.debug(f"Final embedding shape: {embeddings.shape}")
embeddings = layer_norm(embeddings, self.model) embeddings = layer_norm(embeddings)
assert embeddings.dtype == self.fprop_dtype
if last_hid_only: if last_hid_only:
last_step = jnp.maximum(jnp.sum(input_mask.astype(jnp.int32), axis=1) - 1, 0) last_step = jnp.maximum(jnp.sum(input_mask.astype(jnp.int32), axis=1) - 1, 0)