minicpm5-tiny-random

This tiny model is intended for debugging. It is randomly initialized using the configuration adapted from openbmb/MiniCPM5-1B.

Example usage:

from transformers import pipeline

model_id = "yujiepan/minicpm5-tiny-random"
pipe = pipeline(
    "text-generation", model=model_id, device="cuda",
    trust_remote_code=True, max_new_tokens=16,
)
print(pipe("Hello World!"))

Codes to create this repo:

import json

import torch

from huggingface_hub import hf_hub_download
from transformers import (
    AutoConfig,
    AutoModelForCausalLM,
    AutoTokenizer,
    GenerationConfig,
    pipeline,
    set_seed,
)

source_model_id = "openbmb/MiniCPM5-1B"
save_folder = "/tmp/yujiepan/minicpm5-tiny-random"
tokenizer = AutoTokenizer.from_pretrained(
    source_model_id, trust_remote_code=True,
)
tokenizer.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: dict = json.load(f)
config_json.update({
    "hidden_size": 16,
    "intermediate_size": 64,
    "num_attention_heads": 16,
    "num_key_value_heads": 2,
    "head_dim": 32,
    "num_hidden_layers": 2,
})
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,
)

model = AutoModelForCausalLM.from_config(
    config,
    dtype=torch.bfloat16,
    trust_remote_code=True,
)
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, 0.2)
        print(name, p.shape)
model.save_pretrained(save_folder)

Printing the model:

LlamaForCausalLM(
  (model): LlamaModel(
    (embed_tokens): Embedding(130560, 16, padding_idx=1)
    (layers): ModuleList(
      (0-1): 2 x LlamaDecoderLayer(
        (self_attn): LlamaAttention(
          (q_proj): Linear(in_features=16, out_features=512, bias=False)
          (k_proj): Linear(in_features=16, out_features=64, bias=False)
          (v_proj): Linear(in_features=16, out_features=64, bias=False)
          (o_proj): Linear(in_features=512, out_features=16, bias=False)
        )
        (mlp): LlamaMLP(
          (gate_proj): Linear(in_features=16, out_features=64, bias=False)
          (up_proj): Linear(in_features=16, out_features=64, bias=False)
          (down_proj): Linear(in_features=64, out_features=16, bias=False)
          (act_fn): SiLUActivation()
        )
        (input_layernorm): LlamaRMSNorm((16,), eps=1e-06)
        (post_attention_layernorm): LlamaRMSNorm((16,), eps=1e-06)
      )
    )
    (norm): LlamaRMSNorm((16,), eps=1e-06)
    (rotary_emb): LlamaRotaryEmbedding()
  )
  (lm_head): Linear(in_features=16, out_features=130560, bias=False)
)

Test environment:

• torch: 2.10.0+cu128
• transformers: 5.9.0