Spaces:
Sleeping
Sleeping
Oleg Shulyakov
commited on
Commit
·
f0c5d5e
1
Parent(s):
6475b9d
Update _generate_readme
Browse files
app.py
CHANGED
|
@@ -337,51 +337,49 @@ class HuggingFaceModelProcessor:
|
|
| 337 |
|
| 338 |
return new_repo_url
|
| 339 |
|
| 340 |
-
def _generate_readme(self,
|
| 341 |
-
new_repo_id: str, gguf_name: str) -> str:
|
| 342 |
"""Generate README.md for the quantized model."""
|
| 343 |
-
creator = self._get_model_creator(model_id)
|
| 344 |
-
|
| 345 |
-
username = whoami(token)["name"]
|
| 346 |
|
| 347 |
try:
|
| 348 |
-
card = ModelCard.load(model_id, token=token)
|
| 349 |
except:
|
| 350 |
card = ModelCard("")
|
| 351 |
|
| 352 |
if card.data.tags is None:
|
| 353 |
card.data.tags = []
|
| 354 |
card.data.tags.extend(["llama-cpp", "gguf-my-repo"])
|
| 355 |
-
card.data.base_model = model_id
|
| 356 |
|
| 357 |
card.text = dedent(
|
| 358 |
f"""
|
| 359 |
-
# {model_name}
|
| 360 |
**Model creator:** [{creator}](https://huggingface.co/{creator})<br/>
|
| 361 |
-
**Original model**: [{model_id}](https://huggingface.co/{model_id})<br/>
|
| 362 |
**GGUF quantization:** provided by [{username}](https:/huggingface.co/{username}) using `llama.cpp`<br/>
|
| 363 |
## Special thanks
|
| 364 |
🙏 Special thanks to [Georgi Gerganov](https://github.com/ggerganov) and the whole team working on [llama.cpp](https://github.com/ggerganov/llama.cpp/) for making all of this possible.
|
| 365 |
## Use with Ollama
|
| 366 |
```bash
|
| 367 |
-
ollama run "hf.co/{new_repo_id}:<quantization>"
|
| 368 |
```
|
| 369 |
## Use with LM Studio
|
| 370 |
```bash
|
| 371 |
-
lms load "{new_repo_id}"
|
| 372 |
```
|
| 373 |
## Use with llama.cpp CLI
|
| 374 |
```bash
|
| 375 |
-
llama-cli --hf-repo "{new_repo_id}" --hf-file "{
|
| 376 |
```
|
| 377 |
## Use with llama.cpp Server:
|
| 378 |
```bash
|
| 379 |
-
llama-server --hf-repo "{new_repo_id}" --hf-file "{
|
| 380 |
```
|
| 381 |
"""
|
| 382 |
)
|
| 383 |
|
| 384 |
-
readme_path = f"{outdir}/README.md"
|
| 385 |
card.save(readme_path)
|
| 386 |
return readme_path
|
| 387 |
|
|
@@ -422,7 +420,7 @@ llama-server --hf-repo "{new_repo_id}" --hf-file "{gguf_name}" -c 4096
|
|
| 422 |
raise GGUFConverterError(f"Error uploading imatrix.dat: {e}")
|
| 423 |
|
| 424 |
# Upload README.md
|
| 425 |
-
readme_path = self._generate_readme(processing_config
|
| 426 |
self._upload_file(processing_config, readme_path, "README.md")
|
| 427 |
|
| 428 |
print(f"Uploaded successfully with {quant_config.imatrix_method if quant_config.use_imatrix else quant_config.method} option!")
|
|
|
|
| 337 |
|
| 338 |
return new_repo_url
|
| 339 |
|
| 340 |
+
def _generate_readme(self, processing_config :ModelProcessingConfig) -> str:
|
|
|
|
| 341 |
"""Generate README.md for the quantized model."""
|
| 342 |
+
creator = self._get_model_creator(processing_config.model_id)
|
| 343 |
+
username = whoami(processing_config.token)["name"]
|
|
|
|
| 344 |
|
| 345 |
try:
|
| 346 |
+
card = ModelCard.load(processing_config.model_id, token=processing_config.token)
|
| 347 |
except:
|
| 348 |
card = ModelCard("")
|
| 349 |
|
| 350 |
if card.data.tags is None:
|
| 351 |
card.data.tags = []
|
| 352 |
card.data.tags.extend(["llama-cpp", "gguf-my-repo"])
|
| 353 |
+
card.data.base_model = processing_config.model_id
|
| 354 |
|
| 355 |
card.text = dedent(
|
| 356 |
f"""
|
| 357 |
+
# {processing_config.model_name}
|
| 358 |
**Model creator:** [{creator}](https://huggingface.co/{creator})<br/>
|
| 359 |
+
**Original model**: [{processing_config.model_id}](https://huggingface.co/{processing_config.model_id})<br/>
|
| 360 |
**GGUF quantization:** provided by [{username}](https:/huggingface.co/{username}) using `llama.cpp`<br/>
|
| 361 |
## Special thanks
|
| 362 |
🙏 Special thanks to [Georgi Gerganov](https://github.com/ggerganov) and the whole team working on [llama.cpp](https://github.com/ggerganov/llama.cpp/) for making all of this possible.
|
| 363 |
## Use with Ollama
|
| 364 |
```bash
|
| 365 |
+
ollama run "hf.co/{processing_config.new_repo_id}:<quantization>"
|
| 366 |
```
|
| 367 |
## Use with LM Studio
|
| 368 |
```bash
|
| 369 |
+
lms load "{processing_config.new_repo_id}"
|
| 370 |
```
|
| 371 |
## Use with llama.cpp CLI
|
| 372 |
```bash
|
| 373 |
+
llama-cli --hf-repo "{processing_config.new_repo_id}" --hf-file "{processing_config.output_config.filename}" -p "The meaning to life and the universe is"
|
| 374 |
```
|
| 375 |
## Use with llama.cpp Server:
|
| 376 |
```bash
|
| 377 |
+
llama-server --hf-repo "{processing_config.new_repo_id}" --hf-file "{processing_config.output_config.filename}" -c 4096
|
| 378 |
```
|
| 379 |
"""
|
| 380 |
)
|
| 381 |
|
| 382 |
+
readme_path = f"{processing_config.outdir}/README.md"
|
| 383 |
card.save(readme_path)
|
| 384 |
return readme_path
|
| 385 |
|
|
|
|
| 420 |
raise GGUFConverterError(f"Error uploading imatrix.dat: {e}")
|
| 421 |
|
| 422 |
# Upload README.md
|
| 423 |
+
readme_path = self._generate_readme(processing_config)
|
| 424 |
self._upload_file(processing_config, readme_path, "README.md")
|
| 425 |
|
| 426 |
print(f"Uploaded successfully with {quant_config.imatrix_method if quant_config.use_imatrix else quant_config.method} option!")
|