added example usage
Browse files
README.md
CHANGED
|
@@ -18,4 +18,41 @@ It is converted into 8-bit GGUF format from original repository https://huggingf
|
|
| 18 |
|
| 19 |
### Model Sources
|
| 20 |
|
| 21 |
-
**Repository:** https://huggingface.co/defog/sqlcoder-7b-2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
### Model Sources
|
| 20 |
|
| 21 |
+
**Repository:** https://huggingface.co/defog/sqlcoder-7b-2
|
| 22 |
+
|
| 23 |
+
### Example usage
|
| 24 |
+
** with Llamacpp**
|
| 25 |
+
|
| 26 |
+
from langchain_community.llms.llamacpp import LlamaCpp
|
| 27 |
+
from huggingface_hub import hf_hub_download
|
| 28 |
+
|
| 29 |
+
YOUR MODEL DIRECTORY =
|
| 30 |
+
CONTEXT_LENGHT =
|
| 31 |
+
MAX_TOKENS =
|
| 32 |
+
BATCH_SIZE =
|
| 33 |
+
TEMPERATURE =
|
| 34 |
+
GPU_OFFLOAD =
|
| 35 |
+
|
| 36 |
+
def load_model(model_id, model_basename):
|
| 37 |
+
model_path = hf_hub_download(
|
| 38 |
+
repo_id=model_id,
|
| 39 |
+
filename=model_basename,
|
| 40 |
+
resume_download=True,
|
| 41 |
+
cache_dir="YOUR MODEL DIRECTORY",
|
| 42 |
+
)
|
| 43 |
+
kwargs = {
|
| 44 |
+
'model_path': model_path,
|
| 45 |
+
'n_ctx':CONTEXT_LENGHT,
|
| 46 |
+
'max_tokens':MAX_TOKENS,
|
| 47 |
+
'n_batch':BATCH_SIZE,
|
| 48 |
+
'n_gpu_layers':GPU_OFFLOAD,
|
| 49 |
+
'temperature':TEMPERATURE,
|
| 50 |
+
'verbose':True,
|
| 51 |
+
}
|
| 52 |
+
return LlamaCpp(**kwargs)
|
| 53 |
+
|
| 54 |
+
llm = load_model(
|
| 55 |
+
model_id="whoami02/defog-sqlcoder-2-GGUF",
|
| 56 |
+
model_basename="sqlcoder-7b-2.q8_0.gguf",
|
| 57 |
+
)
|
| 58 |
+
|