tiny-random/kormo

This tiny model is intended for debugging. It is randomly initialized using the configuration adapted from KORMo-Team/KORMo-10B-sft.

Example usage:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

model_id = "tiny-random/kormo"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
)
pipe = pipeline('text-generation', model=model, tokenizer=tokenizer, trust_remote_code=True)
print(pipe('Write an article about Artificial Intelligence.'))

Codes to create this repo:

import json
from pathlib import Path

import accelerate
import torch
from huggingface_hub import file_exists, hf_hub_download
from transformers import (
    AutoConfig,
    AutoModelForCausalLM,
    AutoTokenizer,
    GenerationConfig,
    set_seed,
)

source_model_id = "KORMo-Team/KORMo-10B-sft"
save_folder = "/tmp/tiny-random/kormo"

processor = AutoTokenizer.from_pretrained(source_model_id)
processor.save_pretrained(save_folder)

with open(hf_hub_download(source_model_id, filename='config.json', repo_type='model'), 'r', encoding='utf-8') as f:
    config_json = json.load(f)
for k, v in config_json['auto_map'].items():
    config_json['auto_map'][k] = f'{source_model_id}--{v}'

config_json['hidden_size'] = 8
config_json['intermediate_size'] = 64
config_json['num_attention_heads'] = 8
config_json['num_hidden_layers'] = 2
config_json['num_key_value_heads'] = 4

with open(f"{save_folder}/config.json", "w", encoding='utf-8') as f:
    json.dump(config_json, f, indent=2)

config = AutoConfig.from_pretrained(
    save_folder,
    trust_remote_code=True,
)
print(config)

torch.set_default_dtype(torch.bfloat16)
model = AutoModelForCausalLM.from_config(config, trust_remote_code=True)
torch.set_default_dtype(torch.float32)

if file_exists(filename="generation_config.json", repo_id=source_model_id, repo_type='model'):
    model.generation_config = GenerationConfig.from_pretrained(
        source_model_id, trust_remote_code=True,
    )
set_seed(42)
model = model.cpu()
with torch.no_grad():
    for name, p in sorted(model.named_parameters()):
        torch.nn.init.normal_(p, 0, 0.1)
        print(name, p.shape)
model.save_pretrained(save_folder)
print(model)

def modify_automap(path, source_model_id):
    import json
    with open(path, 'r', encoding='utf-8') as f:
        content = json.load(f)
    automap = {}
    if content.get('auto_map', None) is not None:
        for key, value in content.get('auto_map').items():
            if isinstance(value, str):
                value = source_model_id + '--' + value.split('--')[-1]
            else:
                value = [(source_model_id + '--' + v.split('--')[-1]) if '.' in str(v) else v for v in value]
            automap[key] = value
        with open(path, 'w', encoding='utf-8') as f:
            json.dump({**content, 'auto_map': automap}, f, indent=2)

modify_automap(f"{save_folder}/config.json", source_model_id)
# modify_automap(f'{save_folder}/processor_config.json', source_model_id)
# modify_automap(f'{save_folder}/preprocessor_config.json', source_model_id)
# modify_automap(f'{save_folder}/tokenizer_config.json', source_model_id)
for python_file in Path(save_folder).glob('*.py'):
    python_file.unlink()

Printing the model:

KORMoForCausalLM(
  (model): KORMoModel(
    (embed_tokens): Embedding(125184, 8, padding_idx=125032)
    (layers): ModuleList(
      (0-1): 2 x DecoderLayer(
        (self_attn): Attention(
          (q_proj): Linear(in_features=8, out_features=1024, bias=False)
          (k_proj): Linear(in_features=8, out_features=512, bias=False)
          (v_proj): Linear(in_features=8, out_features=512, bias=False)
          (o_proj): Linear(in_features=1024, out_features=8, bias=False)
        )
        (mlp): MLP(
          (gate_proj): Linear(in_features=8, out_features=64, bias=False)
          (up_proj): Linear(in_features=8, out_features=64, bias=False)
          (down_proj): Linear(in_features=64, out_features=8, bias=False)
          (act_fn): SiLU()
        )
        (pre_attention_layernorm): RMSNorm((8,), eps=1e-05)
        (pre_mlp_layernorm): RMSNorm((8,), eps=1e-05)
      )
    )
    (norm): RMSNorm((8,), eps=1e-05)
    (rotary_emb): RotaryEmbedding()
  )
  (lm_head): Linear(in_features=8, out_features=125184, bias=False)
)