git lfs install

git clone https://huggingface.co/ziqingyang/chinese-alpaca-lora-7b

# 不记得有哪些环境的先运行以下命令
conda info -e# 然后激活你需要的环境  我的环境名是py39
conda activate py39

pip install git+https://github.com/huggingface/transformerspip install sentencepiece==0.1.97pip install peft==0.2.0

# Copyright 2022 EleutherAI and The HuggingFace Inc. team. 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 argparse
import gc
import json
import math
import os
import shutil
import warningsimport torchfrom transformers import LlamaConfig, LlamaForCausalLM, LlamaTokenizertry:from transformers import LlamaTokenizerFast
except ImportError as e:warnings.warn(e)warnings.warn("The converted tokenizer will be the `slow` tokenizer. To use the fast, update your `tokenizers` library and re-run the tokenizer conversion")LlamaTokenizerFast = None"""
Sample usage:```
python src/transformers/models/llama/convert_llama_weights_to_hf.py \--input_dir /path/to/downloaded/llama/weights --model_size 7B --output_dir /output/path
```Thereafter, models can be loaded via:```py
from transformers import LlamaForCausalLM, LlamaTokenizermodel = LlamaForCausalLM.from_pretrained("/output/path")
tokenizer = LlamaTokenizer.from_pretrained("/output/path")
```Important note: you need to be able to host the whole model in RAM to execute this script (even if the biggest versions
come in several checkpoints they each contain a part of each weight of the model, so we need to load them all in RAM).
"""INTERMEDIATE_SIZE_MAP = {"7B": 11008,"13B": 13824,"30B": 17920,"65B": 22016,
}
NUM_SHARDS = {"7B": 1,"13B": 2,"30B": 4,"65B": 8,
}def compute_intermediate_size(n):return int(math.ceil(n * 8 / 3) + 255) // 256 * 256def read_json(path):with open(path, "r") as f:return json.load(f)def write_json(text, path):with open(path, "w") as f:json.dump(text, f)def write_model(model_path, input_base_path, model_size):os.makedirs(model_path, exist_ok=True)tmp_model_path = os.path.join(model_path, "tmp")os.makedirs(tmp_model_path, exist_ok=True)params = read_json(os.path.join(input_base_path, "params.json"))num_shards = NUM_SHARDS[model_size]n_layers = params["n_layers"]n_heads = params["n_heads"]n_heads_per_shard = n_heads // num_shardsdim = params["dim"]dims_per_head = dim // n_headsbase = 10000.0inv_freq = 1.0 / (base ** (torch.arange(0, dims_per_head, 2).float() / dims_per_head))# permute for sliced rotarydef permute(w):return w.view(n_heads, dim // n_heads // 2, 2, dim).transpose(1, 2).reshape(dim, dim)print(f"Fetching all parameters from the checkpoint at {input_base_path}.")# Load weightsif model_size == "7B":# Not shared# (The sharded implementation would also work, but this is simpler.)loaded = torch.load(os.path.join(input_base_path, "consolidated.00.pth"), map_location="cpu")else:# Shardedloaded = [torch.load(os.path.join(input_base_path, f"consolidated.{i:02d}.pth"), map_location="cpu")for i in range(num_shards)]param_count = 0index_dict = {"weight_map": {}}for layer_i in range(n_layers):filename = f"pytorch_model-{layer_i + 1}-of-{n_layers + 1}.bin"if model_size == "7B":# Unshardedstate_dict = {f"model.layers.{layer_i}.self_attn.q_proj.weight": permute(loaded[f"layers.{layer_i}.attention.wq.weight"]),f"model.layers.{layer_i}.self_attn.k_proj.weight": permute(loaded[f"layers.{layer_i}.attention.wk.weight"]),f"model.layers.{layer_i}.self_attn.v_proj.weight": loaded[f"layers.{layer_i}.attention.wv.weight"],f"model.layers.{layer_i}.self_attn.o_proj.weight": loaded[f"layers.{layer_i}.attention.wo.weight"],f"model.layers.{layer_i}.mlp.gate_proj.weight": loaded[f"layers.{layer_i}.feed_forward.w1.weight"],f"model.layers.{layer_i}.mlp.down_proj.weight": loaded[f"layers.{layer_i}.feed_forward.w2.weight"],f"model.layers.{layer_i}.mlp.up_proj.weight": loaded[f"layers.{layer_i}.feed_forward.w3.weight"],f"model.layers.{layer_i}.input_layernorm.weight": loaded[f"layers.{layer_i}.attention_norm.weight"],f"model.layers.{layer_i}.post_attention_layernorm.weight": loaded[f"layers.{layer_i}.ffn_norm.weight"],}else:# Sharded# Note that in the 13B checkpoint, not cloning the two following weights will result in the checkpoint# becoming 37GB instead of 26GB for some reason.state_dict = {f"model.layers.{layer_i}.input_layernorm.weight": loaded[0][f"layers.{layer_i}.attention_norm.weight"].clone(),f"model.layers.{layer_i}.post_attention_layernorm.weight": loaded[0][f"layers.{layer_i}.ffn_norm.weight"].clone(),}state_dict[f"model.layers.{layer_i}.self_attn.q_proj.weight"] = permute(torch.cat([loaded[i][f"layers.{layer_i}.attention.wq.weight"].view(n_heads_per_shard, dims_per_head, dim)for i in range(num_shards)],dim=0,).reshape(dim, dim))state_dict[f"model.layers.{layer_i}.self_attn.k_proj.weight"] = permute(torch.cat([loaded[i][f"layers.{layer_i}.attention.wk.weight"].view(n_heads_per_shard, dims_per_head, dim)for i in range(num_shards)],dim=0,).reshape(dim, dim))state_dict[f"model.layers.{layer_i}.self_attn.v_proj.weight"] = torch.cat([loaded[i][f"layers.{layer_i}.attention.wv.weight"].view(n_heads_per_shard, dims_per_head, dim)for i in range(num_shards)],dim=0,).reshape(dim, dim)state_dict[f"model.layers.{layer_i}.self_attn.o_proj.weight"] = torch.cat([loaded[i][f"layers.{layer_i}.attention.wo.weight"] for i in range(num_shards)], dim=1)state_dict[f"model.layers.{layer_i}.mlp.gate_proj.weight"] = torch.cat([loaded[i][f"layers.{layer_i}.feed_forward.w1.weight"] for i in range(num_shards)], dim=0)state_dict[f"model.layers.{layer_i}.mlp.down_proj.weight"] = torch.cat([loaded[i][f"layers.{layer_i}.feed_forward.w2.weight"] for i in range(num_shards)], dim=1)state_dict[f"model.layers.{layer_i}.mlp.up_proj.weight"] = torch.cat([loaded[i][f"layers.{layer_i}.feed_forward.w3.weight"] for i in range(num_shards)], dim=0)state_dict[f"model.layers.{layer_i}.self_attn.rotary_emb.inv_freq"] = inv_freqfor k, v in state_dict.items():index_dict["weight_map"][k] = filenameparam_count += v.numel()torch.save(state_dict, os.path.join(tmp_model_path, filename))filename = f"pytorch_model-{n_layers + 1}-of-{n_layers + 1}.bin"if model_size == "7B":# Unshardedstate_dict = {"model.embed_tokens.weight": loaded["tok_embeddings.weight"],"model.norm.weight": loaded["norm.weight"],"lm_head.weight": loaded["output.weight"],}else:state_dict = {"model.norm.weight": loaded[0]["norm.weight"],"model.embed_tokens.weight": torch.cat([loaded[i]["tok_embeddings.weight"] for i in range(num_shards)], dim=1),"lm_head.weight": torch.cat([loaded[i]["output.weight"] for i in range(num_shards)], dim=0),}for k, v in state_dict.items():index_dict["weight_map"][k] = filenameparam_count += v.numel()torch.save(state_dict, os.path.join(tmp_model_path, filename))# Write configsindex_dict["metadata"] = {"total_size": param_count * 2}write_json(index_dict, os.path.join(tmp_model_path, "pytorch_model.bin.index.json"))config = LlamaConfig(hidden_size=dim,intermediate_size=compute_intermediate_size(dim),num_attention_heads=params["n_heads"],num_hidden_layers=params["n_layers"],rms_norm_eps=params["norm_eps"],)config.save_pretrained(tmp_model_path)# Make space so we can load the model properly now.del state_dictdel loadedgc.collect()print("Loading the checkpoint in a Llama model.")model = LlamaForCausalLM.from_pretrained(tmp_model_path, torch_dtype=torch.float16, low_cpu_mem_usage=True)# Avoid saving this as part of the config.del model.config._name_or_pathprint("Saving in the Transformers format.")model.save_pretrained(model_path)shutil.rmtree(tmp_model_path)def write_tokenizer(tokenizer_path, input_tokenizer_path):# Initialize the tokenizer based on the `spm` modeltokenizer_class = LlamaTokenizer if LlamaTokenizerFast is None else LlamaTokenizerFastprint("Saving a {tokenizer_class} to {tokenizer_path}")tokenizer = tokenizer_class(input_tokenizer_path)tokenizer.save_pretrained(tokenizer_path)def main():parser = argparse.ArgumentParser()parser.add_argument("--input_dir",help="Location of LLaMA weights, which contains tokenizer.model and model folders",)parser.add_argument("--model_size",choices=["7B", "13B", "30B", "65B", "tokenizer_only"],)parser.add_argument("--output_dir",help="Location to write HF model and tokenizer",)args = parser.parse_args()if args.model_size != "tokenizer_only":write_model(model_path=args.output_dir,input_base_path=os.path.join(args.input_dir, args.model_size),model_size=args.model_size,)spm_path = os.path.join(args.input_dir, "tokenizer.model")write_tokenizer(args.output_dir, spm_path)if __name__ == "__main__":main()

python convert_llama_weights_to_hf.py --input_dir path_to_original_llama_root_dir --model_size 7B --output_dir path_to_original_llama_hf_dir

"""
Borrowed and modified from https://github.com/tloen/alpaca-lora
"""import argparse
import os
import json
import gcimport torch
import transformers
import peft
from peft import PeftModelparser = argparse.ArgumentParser()
parser.add_argument('--base_model',default=None,required=True,type=str,help="Please specify a base_model")
parser.add_argument('--lora_model',default=None,required=True,type=str,help="Please specify a lora_model")# deprecated; the script infers the model size from the checkpoint
parser.add_argument('--model_size',default='7B',type=str,help="Size of the LLaMA model",choices=['7B','13B'])parser.add_argument('--offload_dir',default=None,type=str,help="(Optional) Please specify a temp folder for offloading (useful for low-RAM machines). Default None (disable offload).")
parser.add_argument('--output_dir',default='./',type=str)
args = parser.parse_args()assert ("LlamaTokenizer" in transformers._import_structure["models.llama"]
), "LLaMA is now in HuggingFace's main branch.\nPlease reinstall it: pip uninstall transformers && pip install git+https://github.com/huggingface/transformers.git"
from transformers import LlamaTokenizer, LlamaForCausalLMBASE_MODEL = args.base_model
LORA_MODEL = args.lora_model
output_dir = args.output_dirassert (BASE_MODEL
), "Please specify a BASE_MODEL in the script, e.g. 'decapoda-research/llama-7b-hf'"tokenizer = LlamaTokenizer.from_pretrained(LORA_MODEL)
if args.offload_dir is not None:# Load with offloading, which is useful for low-RAM machines.# Note that if you have enough RAM, please use original method instead, as it is faster.base_model = LlamaForCausalLM.from_pretrained(BASE_MODEL,load_in_8bit=False,torch_dtype=torch.float16,offload_folder=args.offload_dir,offload_state_dict=True,low_cpu_mem_usage=True,device_map={"": "cpu"},)
else:# Original method without offloadingbase_model = LlamaForCausalLM.from_pretrained(BASE_MODEL,load_in_8bit=False,torch_dtype=torch.float16,device_map={"": "cpu"},)base_model.resize_token_embeddings(len(tokenizer))
assert base_model.get_input_embeddings().weight.size(0) == len(tokenizer)
tokenizer.save_pretrained(output_dir)
print(f"Extended vocabulary size: {len(tokenizer)}")first_weight = base_model.model.layers[0].self_attn.q_proj.weight
first_weight_old = first_weight.clone()## infer the model size from the checkpoint
emb_to_model_size = {4096 : '7B',5120 : '13B',6656 : '30B',8192 : '65B',
}
embedding_size = base_model.get_input_embeddings().weight.size(1)
model_size = emb_to_model_size[embedding_size]
print(f"Loading LoRA for {model_size} model")lora_model = PeftModel.from_pretrained(base_model,LORA_MODEL,device_map={"": "cpu"},torch_dtype=torch.float16,
)assert torch.allclose(first_weight_old, first_weight)
# merge weights
print(f"Peft version: {peft.__version__}")
print(f"Merging model")
if peft.__version__ > '0.2.0':# merge weights - new merging method from peftlora_model = lora_model.merge_and_unload()
else:# merge weightsfor layer in lora_model.base_model.model.model.layers:if hasattr(layer.self_attn.q_proj,'merge_weights'):layer.self_attn.q_proj.merge_weights = Trueif hasattr(layer.self_attn.v_proj,'merge_weights'):layer.self_attn.v_proj.merge_weights = Trueif hasattr(layer.self_attn.k_proj,'merge_weights'):layer.self_attn.k_proj.merge_weights = Trueif hasattr(layer.self_attn.o_proj,'merge_weights'):layer.self_attn.o_proj.merge_weights = Trueif hasattr(layer.mlp.gate_proj,'merge_weights'):layer.mlp.gate_proj.merge_weights = Trueif hasattr(layer.mlp.down_proj,'merge_weights'):layer.mlp.down_proj.merge_weights = Trueif hasattr(layer.mlp.up_proj,'merge_weights'):layer.mlp.up_proj.merge_weights = Truelora_model.train(False)# did we do anything?
assert not torch.allclose(first_weight_old, first_weight)lora_model_sd = lora_model.state_dict()
del lora_model, base_modelnum_shards_of_models = {'7B': 1, '13B': 2}
params_of_models = {'7B':{"dim": 4096,"multiple_of": 256,"n_heads": 32,"n_layers": 32,"norm_eps": 1e-06,"vocab_size": -1,},'13B':{"dim": 5120,"multiple_of": 256,"n_heads": 40,"n_layers": 40,"norm_eps": 1e-06,"vocab_size": -1,},
}params = params_of_models[model_size]
num_shards = num_shards_of_models[model_size]n_layers = params["n_layers"]
n_heads = params["n_heads"]
dim = params["dim"]
dims_per_head = dim // n_heads
base = 10000.0
inv_freq = 1.0 / (base ** (torch.arange(0, dims_per_head, 2).float() / dims_per_head))def permute(w):return (w.view(n_heads, dim // n_heads // 2, 2, dim).transpose(1, 2).reshape(dim, dim))def unpermute(w):return (w.view(n_heads, 2, dim // n_heads // 2, dim).transpose(1, 2).reshape(dim, dim))def translate_state_dict_key(k):k = k.replace("base_model.model.", "")if k == "model.embed_tokens.weight":return "tok_embeddings.weight"elif k == "model.norm.weight":return "norm.weight"elif k == "lm_head.weight":return "output.weight"elif k.startswith("model.layers."):layer = k.split(".")[2]if k.endswith(".self_attn.q_proj.weight"):return f"layers.{layer}.attention.wq.weight"elif k.endswith(".self_attn.k_proj.weight"):return f"layers.{layer}.attention.wk.weight"elif k.endswith(".self_attn.v_proj.weight"):return f"layers.{layer}.attention.wv.weight"elif k.endswith(".self_attn.o_proj.weight"):return f"layers.{layer}.attention.wo.weight"elif k.endswith(".mlp.gate_proj.weight"):return f"layers.{layer}.feed_forward.w1.weight"elif k.endswith(".mlp.down_proj.weight"):return f"layers.{layer}.feed_forward.w2.weight"elif k.endswith(".mlp.up_proj.weight"):return f"layers.{layer}.feed_forward.w3.weight"elif k.endswith(".input_layernorm.weight"):return f"layers.{layer}.attention_norm.weight"elif k.endswith(".post_attention_layernorm.weight"):return f"layers.{layer}.ffn_norm.weight"elif k.endswith("rotary_emb.inv_freq") or "lora" in k:return Noneelse:print(layer, k)raise NotImplementedErrorelse:print(k)raise NotImplementedErrordef save_shards(lora_model_sd, num_shards: int):# Add the no_grad context managerwith torch.no_grad():if num_shards == 1:new_state_dict = {}for k, v in lora_model_sd.items():new_k = translate_state_dict_key(k)if new_k is not None:if "wq" in new_k or "wk" in new_k:new_state_dict[new_k] = unpermute(v)else:new_state_dict[new_k] = vos.makedirs(output_dir, exist_ok=True)print(f"Saving shard 1 of {num_shards} into {output_dir}/consolidated.00.pth")torch.save(new_state_dict, output_dir + "/consolidated.00.pth")with open(output_dir + "/params.json", "w") as f:json.dump(params, f)else:new_state_dicts = [dict() for _ in range(num_shards)]for k in list(lora_model_sd.keys()):v = lora_model_sd[k]new_k = translate_state_dict_key(k)if new_k is not None:if new_k=='tok_embeddings.weight':print(f"Processing {new_k}")assert v.size(1)%num_shards==0splits = v.split(v.size(1)//num_shards,dim=1)elif new_k=='output.weight':print(f"Processing {new_k}")splits = v.split(v.size(0)//num_shards,dim=0)elif new_k=='norm.weight':print(f"Processing {new_k}")splits = [v] * num_shardselif 'ffn_norm.weight' in new_k:print(f"Processing {new_k}")splits = [v] * num_shardselif 'attention_norm.weight' in new_k:print(f"Processing {new_k}")splits = [v] * num_shardselif 'w1.weight' in new_k:print(f"Processing {new_k}")splits = v.split(v.size(0)//num_shards,dim=0)elif 'w2.weight' in new_k:print(f"Processing {new_k}")splits = v.split(v.size(1)//num_shards,dim=1)elif 'w3.weight' in new_k:print(f"Processing {new_k}")splits = v.split(v.size(0)//num_shards,dim=0)elif 'wo.weight' in new_k:print(f"Processing {new_k}")splits = v.split(v.size(1)//num_shards,dim=1)elif 'wv.weight' in new_k:print(f"Processing {new_k}")splits = v.split(v.size(0)//num_shards,dim=0)elif "wq.weight" in new_k or "wk.weight" in new_k:print(f"Processing {new_k}")v = unpermute(v)splits = v.split(v.size(0)//num_shards,dim=0)else:print(f"Unexpected key {new_k}")raise ValueErrorfor sd,split in zip(new_state_dicts,splits):sd[new_k] = split.clone()del splitdel splitsdel lora_model_sd[k],vgc.collect()    # Effectively enforce garbage collectionos.makedirs(output_dir, exist_ok=True)for i,new_state_dict in enumerate(new_state_dicts):print(f"Saving shard {i+1} of {num_shards} into {output_dir}/consolidated.0{i}.pth")torch.save(new_state_dict, output_dir + f"/consolidated.0{i}.pth")with open(output_dir + "/params.json", "w") as f:print(f"Saving params.json into {output_dir}/params.json")json.dump(params, f)save_shards(lora_model_sd=lora_model_sd, num_shards=num_shards)

python merge_llama_with_chinese_lora.py --base_model path_to_original_llama_hf_dir --lora_model chinese-alpaca-lora-7b --output_dir path_to_output_dir

git clone https://github.com/ggerganov/llama.cpp

cd llama.cpp

cmake . -G "MinGW Makefiles"cmake --build . --config Release

mkdir buildcd buildcmake ..cmake --build . --config Release

python convert-pth-to-ggml.py zh-models/7B/ 1

D:\llama\llama.cpp\bin\quantize.exe ./zh-models/7B/ggml-model-f16.bin ./zh-models/7B/ggml-model-q4_0.bin 2

D:\llama\llama.cpp\bin\main.exe -m zh-models/7B/ggml-model-q4_0.bin --color -f prompts/alpaca.txt -ins -c 2048 --temp 0.2 -n 256 --repeat_penalty 1.3