Upload folder using huggingface_hub

README.md

@@ -0,0 +1,105 @@
+---
+library_name: transformers
+pipeline_tag: text-generation
+inference: true
+widget:
+- text: Hello!
+  example_title: Hello world
+  group: Python
+---
+
+This model is for debugging purposes. It is randomly initialized using the config from [mistralai/Mamba-Codestral-7B-v0.1](https://huggingface.co/mistralai/Mamba-Codestral-7B-v0.1) but with a smaller size. 
+
+Codes:
+```python
+import os
+
+import torch
+
+from huggingface_hub import create_repo, upload_folder
+from transformers import (
+    AutoModelForCausalLM,
+    AutoTokenizer,
+    GenerationConfig,
+    Mamba2Config,
+    pipeline,
+    set_seed,
+)
+
+model_id = "mistralai/Mamba-Codestral-7B-v0.1"
+repo_id = "yujiepan/mamba2-codestral-v0.1-tiny-random"
+save_path = f"/tmp/{repo_id}"
+
+os.system(f'rm -rf {save_path}')
+
+config = Mamba2Config.from_pretrained(model_id)
+config.use_cache = True
+config.num_hidden_layers = 2
+config.num_heads = 8
+config.head_dim = 4
+config.hidden_size = 8
+config.expand = 4
+config.intermediate_size = 32
+config.state_size = 8
+config.n_groups = 2
+
+assert config.intermediate_size == \
+    config.hidden_size * config.expand == config.num_heads * config.head_dim
+assert config.num_heads // config.n_groups > 0
+assert config.num_heads % 8 == 0
+
+tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
+tokenizer.save_pretrained(save_path)
+
+model = AutoModelForCausalLM.from_config(
+    config, torch_dtype=torch.bfloat16,
+    trust_remote_code=True,
+)
+model.generation_config = GenerationConfig.from_pretrained(
+    model_id,
+    trust_remote_code=True,
+)
+
+set_seed(42)
+with torch.no_grad():
+    for name, p in sorted(model.named_parameters()):
+        print(name, p.shape)
+        torch.nn.init.uniform_(p, -0.5, 0.5)
+
+model.save_pretrained(save_path)
+
+pipe = pipeline(
+    "text-generation",
+    model=save_path,
+    device="cuda",
+    trust_remote_code=True,
+    max_new_tokens=20,
+)
+print(pipe("Hello World!"))
+
+with open(__file__, 'r') as f:
+    codes = f.read()
+with open(f'{save_path}/README.md', 'w') as f:
+    f.write(
+        f'''---
+library_name: transformers
+pipeline_tag: text-generation
+inference: true
+widget:
+- text: Hello!
+  example_title: Hello world
+  group: Python
+---
+
+This model is for debugging purposes. It is randomly initialized using the config from [{model_id}](https://huggingface.co/{model_id}) but with a smaller size. 
+
+Codes:
+```python
+{codes}
+```'''
+    )
+
+create_repo(repo_id, exist_ok=True)
+upload_folder(repo_id=repo_id, folder_path=save_path, repo_type='model')
+
+```

config.json

@@ -0,0 +1,44 @@
+{
+  "_name_or_path": "/raid/pablo/codestral-hf-good/",
+  "architectures": [
+    "Mamba2ForCausalLM"
+  ],
+  "bos_token_id": 0,
+  "chunk_size": 256,
+  "conv_kernel": 4,
+  "eos_token_id": 0,
+  "expand": 4,
+  "head_dim": 4,
+  "hidden_act": "silu",
+  "hidden_size": 8,
+  "initializer_range": 0.1,
+  "intermediate_size": 32,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "mamba2",
+  "n_groups": 2,
+  "norm_before_gate": true,
+  "num_heads": 8,
+  "num_hidden_layers": 2,
+  "pad_token_id": 0,
+  "rescale_prenorm_residual": false,
+  "residual_in_fp32": true,
+  "rms_norm": true,
+  "state_size": 8,
+  "tie_word_embeddings": false,
+  "time_step_floor": 0.0001,
+  "time_step_init_scheme": "random",
+  "time_step_limit": [
+    0.0,
+    Infinity
+  ],
+  "time_step_max": 0.1,
+  "time_step_min": 0.001,
+  "time_step_rank": 256,
+  "time_step_scale": 1.0,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.45.2",
+  "use_bias": false,
+  "use_cache": true,
+  "use_conv_bias": true,
+  "vocab_size": 32768
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 0,
+  "eos_token_id": 2,
+  "pad_token_id": 1,
+  "transformers_version": "4.45.2"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bf16d21b31bed6b3d74761a73c2ee17897fba814c780b7c8f1c228eb8789c90
+size 1056544

special_tokens_map.json

@@ -0,0 +1,23 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59f95e28944c062244741268596badc900df86c7f5ded05088d2da22a7379e06
+size 587583