feat(dsv3):Runnable N1C8 configs

paddleformers/examples/deepseek_v3/config/__init__.py

@@ -0,0 +1,100 @@
+# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+import sys
+from typing import TYPE_CHECKING
+
+from paddleformers.utils.lazy_import import _LazyModule
+
+import_structure = {
+    "configuration": ["DeepseekV2FastConfig"],
+    "modeling": [
+        "masked_fill",
+        "DeepseekV2Attention",
+        "MoEGate",
+        "FakeGate",
+        "DeepseekV2ForCausalLM",
+        "_make_causal_mask",
+        "is_casual_mask",
+        "DeepseekV2MoE",
+        "DeepseekV2MoEFlexToken",
+        "scaled_dot_product_attention",
+        "DeepseekV2RotaryEmbedding",
+        "rotate_half",
+        "DeepseekV2MTPLayer",
+        "DeepseekV2RMSNorm",
+        "DeepseekV2YarnRotaryEmbedding",
+        "parallel_matmul",
+        "DeepseekV2PretrainedModel",
+        "AddAuxiliaryLoss",
+        "apply_rotary_pos_emb",
+        "assign_kv_heads",
+        "DeepseekV2ForSequenceClassification",
+        "_expand_2d_mask",
+        "DeepseekV2ModelFast",
+        "repeat_kv",
+        "yarn_find_correction_dim",
+        "yarn_linear_ramp_mask",
+        "DeepseekV2DynamicNTKScalingRotaryEmbedding",
+        "DeepseekV2MLP",
+        "yarn_get_mscale",
+        "DeepseekV2LMHead",
+        "DeepseekV2DecoderLayer",
+        "DeepseekV2PretrainingCriterionFast",
+        "yarn_find_correction_range",
+        "get_triangle_upper_mask",
+        "DeepseekV2LinearScalingRotaryEmbedding",
+        "set_global_step",
+        "get_global_step",
+    ],
+    "modeling_auto": [
+        "DeepseekV2LMHeadAuto",
+        "DeepseekV2ForCausalLMAuto",
+        "DeepseekV2ModelAuto",
+        "DeepseekV2PretrainedModelAuto",
+    ],
+    "modeling_pp": ["DeepseekV2ForCausalLMPipe"],
+    "mfu_utils": ["DeepSeekProjection"],
+    "kernel": [
+        "act_quant",
+        "weight_dequant",
+        "fp8_gemm",
+        "weight_dequant_kernel",
+        "act_quant_kernel",
+        "fp8_gemm_kernel",
+    ],
+    "tokenizer_fast": ["DeepseekTokenizerFast"],
+    "fp8_linear": [
+        "Linear",
+        "ColumnParallelLinear",
+        "RowParallelLinear",
+        "ColumnSequenceParallelLinear",
+        "RowSequenceParallelLinear",
+    ],
+}
+
+if TYPE_CHECKING:
+    from .configuration import *
+    from .modeling import *
+    from .modeling_auto import *
+    from .modeling_pp import *
+    from .tokenizer_fast import *
+else:
+    sys.modules[__name__] = _LazyModule(
+        __name__,
+        globals()["__file__"],
+        import_structure,
+        module_spec=__spec__,
+    )

paddleformers/examples/deepseek_v3/config/config.json

@@ -0,0 +1,80 @@
+{
+    "architectures": [
+      "DeepseekV2ForCausalLM"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "auto_map": {
+      "AutoConfig": "DeepseekV2FastConfig",
+      "AutoModel": "DeepseekV2ModelFast",
+      "AutoModelForCausalLM": "DeepseekV2ForCausalLM"
+    },
+    "aux_loss_alpha": 0.001,
+    "bos_token_id": 0,
+    "eos_token_id": 1,
+    "ep_size": 1,
+    "first_k_dense_replace": 3,
+    "hidden_act": "silu",
+    "hidden_size": 7168,
+    "initializer_range": 0.02,
+    "intermediate_size": 18432,
+    "kv_lora_rank": 512,
+    "max_position_embeddings": 163840,
+    "model_type": "deepseek_v3",
+    "moe_intermediate_size": 2048,
+    "moe_layer_freq": 1,
+    "n_group": 8,
+    "n_routed_experts": 8,
+    "n_shared_experts": 1,
+    "norm_topk_prob": true,
+    "num_attention_heads": 128,
+    "num_experts_per_tok": 8,
+    "num_hidden_layers": 15,
+    "num_key_value_heads": 128,
+    "num_nextn_predict_layers": 1,
+    "pretraining_tp": 1,
+    "q_lora_rank": 1536,
+    "qk_nope_head_dim": 128,
+    "qk_rope_head_dim": 64,
+    "rms_norm_eps": 1e-06,
+    "rope_scaling": {
+      "beta_fast": 32,
+      "beta_slow": 1,
+      "factor": 40,
+      "mscale": 1.0,
+      "mscale_all_dim": 1.0,
+      "original_max_position_embeddings": 4096,
+      "type": "yarn"
+    },
+    "rope_theta": 10000,
+    "routed_scaling_factor": 2.5,
+    "scoring_func": "sigmoid",
+    "seq_aux": true,
+    "tie_word_embeddings": false,
+    "topk_group": 4,
+    "topk_method": "noaux_tc",
+    "dtype": "bfloat16",
+    "transformers_version": "4.33.1",
+    "use_cache": true,
+    "v_head_dim": 128,
+    "vocab_size": 129280,
+    "using_flex_token": true,
+    "using_fake_gate": true,
+    "use_fused_rms_norm": true,
+    "fuse_attention_ffn": true,
+    "use_fused_rope": true,
+    "token_drop_steps": 0,
+    "recompute_fwd_gate_up": true,
+    "adaptive_remained_O1_recompute_ratio": 0.3,
+    "using_post_norm_recompute": true,
+    "is_split_group_gemm": false,
+    "use_dualpipev": true,
+    "send_mtp_embed": true,
+    "offline_quant_expert_weight": false,
+    "clear_origin_weight_when_offline_quant": false,
+    "dsv3_use_fp8_gemm": true,
+    "dsv3_use_atten_recompute": true,
+    "use_ds_gemm": false,
+    "dsv3_use_fp8_dispatch": true,
+    "fa_version": 3
+  }