mirror of
https://github.com/FlashML-org/FreeToken.git
synced 2026-11-23 02:00:02 +08:00
* refactor(quant): config, scheme and method layers for checkpoint quantization QuantConfig parses the checkpoint's quantization_config into a QuantScheme per projection; every Linear*, ParallelLMHead and MoE layer takes quant_config and a prefix, declares its tensors through a QuantMethod and selects its kernel at construction. MoE kernels own the expert bank layout, pack and apply, so the offload cache, the FTW converter and the CPU executor follow one layout. - --quant-backend layer[.kind]=name selects kernels (--nvfp4-backend forwards to it) - --moe-strategy replaces --moe-backend; the old spelling stays as an alias - --moe-cpu-layers auto for Windows/WSL pin budgets; --moe-cache-auto respects the kernel's slot limit - removes the per-family quantized layer classes, bank providers, format env switches and GLM runtime fp8
98 lines
3.8 KiB
Python
98 lines
3.8 KiB
Python
from __future__ import annotations
|
|
|
|
from typing import TYPE_CHECKING, Tuple
|
|
|
|
import torch
|
|
from freetoken.core import get_global_ctx
|
|
from freetoken.layers import BaseOP, OPList, ParallelLMHead, RMSNormFused, VocabParallelEmbedding
|
|
from freetoken.models.blocks import BaseLLMModel
|
|
from freetoken.utils import nvtx_annotate
|
|
|
|
from .attention import GlmMoeDsaAttention
|
|
from .mlp import GlmDsaGatedMLP
|
|
from .moe import GlmMoeDsaSparseBlock
|
|
|
|
if TYPE_CHECKING:
|
|
from freetoken.models.config import ModelConfig
|
|
|
|
|
|
class GlmMoeDsaDecoderLayer(BaseOP):
|
|
def __init__(self, config: ModelConfig, layer_id: int, *, prefix: str = ""):
|
|
self.self_attn = GlmMoeDsaAttention(config, layer_id, prefix=f"{prefix}.self_attn")
|
|
if layer_id >= config.first_k_dense_replace:
|
|
self.mlp: BaseOP = GlmMoeDsaSparseBlock(config, layer_id, prefix=f"{prefix}.mlp")
|
|
else:
|
|
self.mlp = GlmDsaGatedMLP(
|
|
config.hidden_size, config.intermediate_size,
|
|
quant_config=config.quant, prefix=f"{prefix}.mlp",
|
|
)
|
|
self.input_layernorm = RMSNormFused(size=config.hidden_size, eps=config.rms_norm_eps)
|
|
self.post_attention_layernorm = RMSNormFused(
|
|
size=config.hidden_size, eps=config.rms_norm_eps
|
|
)
|
|
self._layer_id = layer_id
|
|
|
|
@nvtx_annotate("Layer_{}", layer_id_field="_layer_id")
|
|
def forward(
|
|
self, x: torch.Tensor, residual: torch.Tensor | None = None
|
|
) -> Tuple[torch.Tensor, torch.Tensor]:
|
|
x, residual = self.input_layernorm.forward(x, residual)
|
|
x = self.self_attn.forward(x)
|
|
x, residual = self.post_attention_layernorm.forward(x, residual)
|
|
x = self.mlp.forward(x)
|
|
return x, residual
|
|
|
|
|
|
class GlmMoeDsaModel(BaseOP):
|
|
def __init__(self, config: ModelConfig, *, prefix: str = "model"):
|
|
self.embed_tokens = VocabParallelEmbedding(
|
|
num_embeddings=config.vocab_size,
|
|
embedding_dim=config.hidden_size,
|
|
)
|
|
self.layers = OPList(
|
|
[
|
|
GlmMoeDsaDecoderLayer(config, layer_id, prefix=f"{prefix}.layers.{layer_id}")
|
|
for layer_id in range(config.num_layers)
|
|
]
|
|
)
|
|
self.norm = RMSNormFused(size=config.hidden_size, eps=config.rms_norm_eps)
|
|
|
|
def forward(self, input_ids: torch.Tensor) -> torch.Tensor:
|
|
x = self.embed_tokens.forward(input_ids)
|
|
residual: torch.Tensor | None = None
|
|
for layer in self.layers.op_list:
|
|
x, residual = layer.forward(x, residual)
|
|
return self.norm.forward(x, residual)[0]
|
|
|
|
|
|
class GlmMoeDsaForCausalLM(BaseLLMModel):
|
|
def __init__(self, config: ModelConfig):
|
|
self.model = GlmMoeDsaModel(config)
|
|
self.lm_head = ParallelLMHead(
|
|
num_embeddings=config.vocab_size,
|
|
embedding_dim=config.hidden_size,
|
|
tie_word_embeddings=config.tie_word_embeddings,
|
|
tied_embedding=self.model.embed_tokens if config.tie_word_embeddings else None,
|
|
quant_config=config.quant,
|
|
prefix="lm_head",
|
|
)
|
|
super().__init__()
|
|
|
|
def prepare_for_runtime(self) -> None:
|
|
"""Post-load, pre-KV-sizing hook (engine calls it before the pool family's solve_num_pages):
|
|
materialize every layer's bmm-ready kv_b split and free the checkpoint-layout
|
|
originals, so the ~2.2 GiB repack is measured by the sizing pass instead of
|
|
overcommitting the KV budget on the first forward (gpt_oss precedent)."""
|
|
import torch
|
|
|
|
for layer in self.model.layers.op_list:
|
|
layer.self_attn.prepare_for_runtime()
|
|
torch.cuda.empty_cache()
|
|
|
|
def forward(self) -> torch.Tensor:
|
|
output = self.model.forward(get_global_ctx().batch.input_ids)
|
|
return self.lm_head.forward(output)
|
|
|
|
|
|
__all__ = ["GlmMoeDsaForCausalLM"]
|