basically same converter but with different config

Author: haeggee

Diff for: examples/xglm/convert_dense2langmoe.py

@@ -0,0 +1,197 @@
+"""
+Converts a nanotron model to HF format
+Command:
+    torchrun --nproc-per-node=1 convert_dense2moe.py --checkpoint-path=nanotron_weights --save-path=nanotron_moe_weights
+"""
+
+import dataclasses
+import json
+import warnings
+from argparse import ArgumentParser
+from pathlib import Path
+from typing import Optional
+
+from torch import nn
+import torch
+import nanotron
+from nanotron.config.models_config import GPT3Config, GPT3LangMoEConfig
+from nanotron.models.gpt3 import GPT3ForTraining, GPTBlock
+from nanotron.models.gpt3_langmoe import GPT3LangMoEForTraining, GPT3LangMoEBlock
+from nanotron.trainer import mark_tied_parameters
+
+from convert_utils import convert_generic, create_nt_model
+
+
+def convert_config(config: GPT3Config, num_experts=8, num_languages=32, language_embedding_size=128) -> GPT3LangMoEConfig:
+    return GPT3LangMoEConfig(
+        **config.__dict__,
+        is_moe=True,
+        moe_num_experts=num_experts,
+        num_experts_per_tok=min(2, num_experts),  # arbitrarily chosen
+        moe_loss_weight=0.01,  # arbitrarily chosen
+        moe_z_loss_weight=0.001,  # arbitrarily chosen
+        moe_glu=False,
+        num_languages=num_languages,
+        language_embedding_size=language_embedding_size,
+    )
+
+
+def convert_dense_to_moe(ff_moe: nn.Module, dense_ff: nn.Module, num_experts: int):
+    with torch.no_grad():
+        # only copy the weight matrix and repeat it n_expert times
+        weight_1 = dense_ff.c_fc.weight.clone()
+        if num_experts == 1:
+            ff_moe.experts.mlp.w1.module.weight.data = weight_1.contiguous()
+        else:
+            # [intermediate_size, hidden_size] -> [hidden_size, intermediate_size * n_experts]
+            weight_1 = weight_1.T
+            ff_moe.experts.mlp.w1.module.weight.data = weight_1.repeat(1, num_experts)
+
+        weight_2 = dense_ff.c_proj.weight.clone()
+        if num_experts == 1: # just a specific case for 1 expert
+            ff_moe.experts.mlp.w2.module.weight.data = weight_2.contiguous()
+        else: 
+            # [hidden_size, intermediate_size] -> [intermediate_size * n_experts, hidden_size]
+            weight_2 = weight_2.T
+            ff_moe.experts.mlp.w2.module.weight.data = weight_2.repeat(num_experts, 1)
+
+        # # -- could add bias only for 2nd layer, because that works with the MegaBlocks MoE implementation
+        # # -- but won't make a big difference?
+        # ff_moe.experts.bias.copy_(dense_ff.c_proj.bias)
+
+        # init gating randomly
+        nn.init.normal_(ff_moe.gate.layer.weight, mean=0.0, std=0.02)
+
+
+def convert_decoder(block_moe: GPT3LangMoEBlock, block_nt: GPTBlock, num_experts: int):
+    convert_generic(block_moe.ln_1, block_nt.ln_1)
+    convert_generic(block_moe.attn, block_nt.attn)
+    convert_generic(block_moe.ln_2, block_nt.ln_2)
+    convert_dense_to_moe(block_moe.ff, block_nt.ff, num_experts)
+
+
+def convert(
+    model_moe: GPT3LangMoEForTraining, model_dense: GPT3ForTraining, num_experts: int
+):
+    convert_generic(
+        model_moe.model.token_embeddings.pp_block.token_embedding,
+        model_dense.model.token_embeddings.pp_block.token_embedding,
+    )
+    # init laguage embedding randomly
+    nn.init.normal_(model_moe.model.language_embeddings.pp_block.language_embedding.weight, mean=0.0, std=0.02)
+    for layer_moe, layer_nt in zip(model_moe.model.decoder, model_dense.model.decoder):
+        convert_decoder(layer_moe.pp_block, layer_nt.pp_block, num_experts)
+    convert_generic(
+        model_moe.model.final_layer_norm.pp_block,
+        model_dense.model.final_layer_norm.pp_block,
+    )
+    convert_generic(
+        model_moe.model.lm_head.pp_block, model_dense.model.lm_head.pp_block
+    )
+
+
+def create_nt_moe_model(
+    model_config: Optional[GPT3Config] = None,
+    device: torch.device = torch.device("cuda"),
+    dtype: torch.dtype = torch.bfloat16,
+    checkpoint_path: Optional[Path] = None,
+):
+
+    if model_config is None:
+        assert checkpoint_path is not None
+        with open(checkpoint_path / "model_config.json") as f:
+            model_config = GPT3LangMoEConfig(**json.load(f))
+
+    parallel_config = nanotron.config.ParallelismArgs(dp=1, pp=1, tp=1)
+    parallel_context = nanotron.parallel.ParallelContext(
+        data_parallel_size=parallel_config.dp,
+        pipeline_parallel_size=parallel_config.pp,
+        tensor_parallel_size=parallel_config.tp,
+    )
+    model_nt = nanotron.models.build_model(
+        model_builder=lambda: GPT3LangMoEForTraining(
+            config=model_config,
+            parallel_context=parallel_context,
+            parallel_config=parallel_config,
+            random_states=None,
+        ),
+        parallel_context=parallel_context,
+        dtype=dtype,
+        device=device,
+    )
+    mark_tied_parameters(model=model_nt, parallel_context=parallel_context)
+
+    if checkpoint_path is not None:
+        nanotron.serialize.load_weights(
+            model=model_nt,
+            parallel_context=parallel_context,
+            root_folder=checkpoint_path,
+        )
+
+    return model_nt
+
+
+def main(
+    checkpoint_path: Path,
+    save_path: Path,
+    num_experts: int,
+    num_languages: int,
+    language_embedding_size: int,
+):
+    # Load nanotron model.
+    model_dense = create_nt_model(checkpoint_path=checkpoint_path)
+
+    # Init moe model.
+    model_config_moe = convert_config(model_dense.config, num_experts, num_languages, language_embedding_size)
+    model_moe = create_nt_moe_model(model_config=model_config_moe)
+
+    convert(model_moe, model_dense, num_experts)
+    nanotron.serialize.save_weights(
+        model=model_moe,
+        parallel_context=model_moe.parallel_context,
+        root_folder=save_path,
+    )
+    with open(save_path / "model_config.json", "w+") as f:
+        json.dump(dataclasses.asdict(model_config_moe), f)
+    print(f"Model saved to {save_path}")
+
+
+if __name__ == "__main__":
+    # fix all random seeds
+    torch.manual_seed(0)
+    torch.cuda.manual_seed(0)
+    torch.cuda.manual_seed_all(0)
+    torch.backends.cudnn.deterministic = True
+    parser = ArgumentParser(description="Convert dense weights to moe format")
+    parser.add_argument(
+        "--checkpoint-path",
+        type=Path,
+        default="checkpoints/xglm-7.5B",
+        help="Path to the nanotron dense checkpoint",
+    )
+    parser.add_argument(
+        "--save-path",
+        type=Path,
+        default="checkpoints/xglm-moe-7.5B",
+        help="Path to save the nanotron moe model",
+    )
+    parser.add_argument(
+        "--num-experts",
+        type=int,
+        default=8,
+        help="Number of experts in the MoE model (duplicates of MLP layer)",
+    )
+    parser.add_argument(
+        "--num-languages",
+        type=int,
+        default=32,
+        help="Number of languages for the language embedding",
+    )
+    parser.add_argument(
+        "--language-embedding-size",
+        type=int,
+        default=128,
+        help="Size of the language embedding",
+    )
+    args = parser.parse_args()
+    main(args.checkpoint_path, args.save_path, args.num_experts, args.num_languages, args.language_embedding_size)