| --- |
| license: mit |
| --- |
| |
|
|
| from pathlib import Path |
| from huggingface_hub import HfApi |
| |
| repo_id = "rekhi/SLM" |
|
|
| readme = """--- |
| library_name: pytorch |
| license: mit |
| tags: |
| - pytorch |
| - custom-model |
| - slm |
| pipeline_tag: text-generation |
| --- |
|
|
| # SLM |
|
|
| Lightweight SLM weights (`slm_model.pt`). |
| > Replace this section with: architecture, training data, intended use, and limitations. |
|
|
| ## Usage |
|
|
| ```python |
| import torch |
| from huggingface_hub import hf_hub_download |
| |
| # Download the file |
| ckpt_path = hf_hub_download(repo_id="rekhi/SLM", filename="slm_model.pt") |
| |
| # 1) If this is a full torch model (saved with torch.save(model)): |
| model = torch.load(ckpt_path, map_location="cpu") |
| model.eval() |
| |
| # 2) If this is a state_dict (saved with torch.save(model.state_dict())): |
| # Define your model class exactly as used in training, then: |
| # model = MyModel(**config) |
| # model.load_state_dict(torch.load(ckpt_path, map_location="cpu")) |
| # model.eval() |
| |
| |
| |
