mirror of
https://github.com/kvcache-ai/ktransformers.git
synced 2026-05-05 15:40:13 +00:00
57d14d22bc
* refactor: move legacy code to archive/ directory - Moved ktransformers, csrc, third_party, merge_tensors to archive/ - Moved build scripts and configurations to archive/ - Kept kt-kernel, KT-SFT, doc, and README files in root - Preserved complete git history for all moved files * refactor: restructure repository to focus on kt-kernel and KT-SFT modules * fix README * fix README * fix README * fix README * docs: add performance benchmarks to kt-kernel section Add comprehensive performance data for kt-kernel to match KT-SFT's presentation: - AMX kernel optimization: 21.3 TFLOPS (3.9× faster than PyTorch) - Prefill phase: up to 20× speedup vs baseline - Decode phase: up to 4× speedup - NUMA optimization: up to 63% throughput improvement - Multi-GPU (8×L20): 227.85 tokens/s total throughput with DeepSeek-R1 FP8 Source: https://lmsys.org/blog/2025-10-22-KTransformers/ This provides users with concrete performance metrics for both core modules, making it easier to understand the capabilities of each component. * refactor: improve kt-kernel performance data with specific hardware and models Replace generic performance descriptions with concrete benchmarks: - Specify exact hardware: 8×L20 GPU + Xeon Gold 6454S, Single/Dual-socket Xeon + AMX - Include specific models: DeepSeek-R1-0528 (FP8), DeepSeek-V3 (671B) - Show detailed metrics: total throughput, output throughput, concurrency details - Match KT-SFT presentation style for consistency This provides users with actionable performance data they can use to evaluate hardware requirements and expected performance for their use cases. * fix README * docs: clean up performance table and improve formatting * add pic for README * refactor: simplify .gitmodules and backup legacy submodules - Remove 7 legacy submodules from root .gitmodules (archive/third_party/*) - Keep only 2 active submodules for kt-kernel (llama.cpp, pybind11) - Backup complete .gitmodules to archive/.gitmodules - Add documentation in archive/README.md for researchers who need legacy submodules This reduces initial clone size by ~500MB and avoids downloading unused dependencies. * refactor: move doc/ back to root directory Keep documentation in root for easier access and maintenance. * refactor: consolidate all images to doc/assets/ - Move kt-kernel/assets/heterogeneous_computing.png to doc/assets/ - Remove KT-SFT/assets/ (images already in doc/assets/) - Update KT-SFT/README.md image references to ../doc/assets/ - Eliminates ~7.9MB image duplication - Centralizes all documentation assets in one location * fix pic path for README
70 lines
No EOL
3.4 KiB
Python
70 lines
No EOL
3.4 KiB
Python
|
|
from ktransformers.operators.base_operator import BaseInjectedModule
|
|
from ktransformers.util.custom_loader import GGUFLoader
|
|
from transformers import PretrainedConfig
|
|
import torch.nn as nn
|
|
from ktransformers.models.modeling_deepseek_v3 import DeepseekV3MLP
|
|
from ktransformers.models.modeling_qwen2_moe import Qwen2MoeMLP
|
|
from ktransformers.models.modeling_smallthinker import SmallthinkerDenseMlpBlock
|
|
from ktransformers.models.modeling_glm4_moe import Glm4MoeMLP
|
|
class kDeepseekV3MLP(DeepseekV3MLP, BaseInjectedModule):
|
|
def __init__(self,
|
|
key: str,
|
|
gguf_loader : GGUFLoader,
|
|
config: PretrainedConfig,
|
|
orig_module: nn.Module,
|
|
prefill_device: str = "cuda",
|
|
generate_device: str = "cuda",
|
|
**kwargs):
|
|
BaseInjectedModule.__init__(self, key, gguf_loader, config, orig_module, prefill_device, **kwargs)
|
|
self.orig_module.__init__(orig_module.config,
|
|
orig_module.hidden_size, orig_module.intermediate_size)
|
|
def forward(self, x, bsz_tensor):
|
|
down_proj = self.down_proj(self.act_fn(self.gate_proj(x, bsz_tensor)) * self.up_proj(x, bsz_tensor), bsz_tensor)
|
|
return down_proj
|
|
class KQwen2MoeMLP(Qwen2MoeMLP, BaseInjectedModule):
|
|
def __init__(self,
|
|
key: str,
|
|
gguf_loader : GGUFLoader,
|
|
config: PretrainedConfig,
|
|
orig_module: nn.Module,
|
|
prefill_device: str = "cuda",
|
|
generate_device: str = "cuda",
|
|
**kwargs):
|
|
BaseInjectedModule.__init__(self, key, gguf_loader, config, orig_module, prefill_device, **kwargs)
|
|
self.orig_module.__init__(orig_module.config,
|
|
orig_module.intermediate_size)
|
|
def forward(self, x, bsz_tensor):
|
|
down_proj = self.down_proj(self.act_fn(self.gate_proj(x, bsz_tensor)) * self.up_proj(x, bsz_tensor), bsz_tensor)
|
|
return down_proj
|
|
|
|
|
|
class KSmallthinkerDenseMlpBlock(SmallthinkerDenseMlpBlock, BaseInjectedModule):
|
|
def __init__(self,
|
|
key: str,
|
|
gguf_loader : GGUFLoader,
|
|
config: PretrainedConfig,
|
|
orig_module: nn.Module,
|
|
prefill_device: str = "cuda",
|
|
generate_device: str = "cuda",
|
|
**kwargs):
|
|
BaseInjectedModule.__init__(self, key, gguf_loader, config, orig_module, prefill_device, **kwargs)
|
|
self.orig_module.__init__(orig_module.config)
|
|
def forward(self, x, bsz_tensor):
|
|
down_proj = self.down(nn.functional.relu(self.gate(x, bsz_tensor)) * self.up(x, bsz_tensor), bsz_tensor)
|
|
return down_proj
|
|
|
|
class KGlm4MoeMLP(Glm4MoeMLP, BaseInjectedModule):
|
|
def __init__(self,
|
|
key: str,
|
|
gguf_loader : GGUFLoader,
|
|
config: PretrainedConfig,
|
|
orig_module: nn.Module,
|
|
prefill_device: str = "cuda",
|
|
generate_device: str = "cuda",
|
|
**kwargs):
|
|
BaseInjectedModule.__init__(self, key, gguf_loader, config, orig_module, prefill_device, **kwargs)
|
|
self.orig_module.__init__(orig_module.config, orig_module.hidden_size, orig_module.intermediate_size)
|
|
def forward(self, x, bsz_tensor):
|
|
down_proj = self.down_proj(self.act_fn(self.gate_proj(x, bsz_tensor)) * self.up_proj(x, bsz_tensor), bsz_tensor)
|
|
return down_proj |