Sync from GitHub via hub-sync
Browse files- README.md +2 -0
- app/backend/minicpm_transformers.py +9 -1
README.md
CHANGED
|
@@ -63,6 +63,8 @@ SMOLNALYSIS_MINICPM_CKAN_RETRIEVAL_TEMPERATURE=0
|
|
| 63 |
SMOLNALYSIS_MINICPM_MAX_NEW_TOKENS=384
|
| 64 |
```
|
| 65 |
|
|
|
|
|
|
|
| 66 |
Upload the trained adapter from a machine with `HF_TOKEN` set:
|
| 67 |
|
| 68 |
```bash
|
|
|
|
| 63 |
SMOLNALYSIS_MINICPM_MAX_NEW_TOKENS=384
|
| 64 |
```
|
| 65 |
|
| 66 |
+
`SMOLNALYSIS_MINICPM_CKAN_RETRIEVAL_ADAPTER_REPO_ID` defaults to the repo above, so the Space uses the uploaded CKAN adapter automatically unless this value is overridden.
|
| 67 |
+
|
| 68 |
Upload the trained adapter from a machine with `HF_TOKEN` set:
|
| 69 |
|
| 70 |
```bash
|
app/backend/minicpm_transformers.py
CHANGED
|
@@ -27,6 +27,10 @@ DEFAULT_MODEL_ID = os.getenv("SMOLNALYSIS_MINICPM_TRANSFORMERS_MODEL_ID", os.get
|
|
| 27 |
DEFAULT_MAX_NEW_TOKENS = int(os.getenv("SMOLNALYSIS_MINICPM_MAX_NEW_TOKENS", os.getenv("MAX_TOKENS", "384")))
|
| 28 |
DEFAULT_TEMPERATURE = float(os.getenv("SMOLNALYSIS_MINICPM_TEMPERATURE", os.getenv("TEMPERATURE", "0.7")))
|
| 29 |
DEFAULT_TOP_P = float(os.getenv("SMOLNALYSIS_MINICPM_TOP_P", os.getenv("TOP_P", "0.95")))
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
ZERO_GPU_DURATION_SECONDS = int(os.getenv("SMOLNALYSIS_MINICPM_ZEROGPU_DURATION_SECONDS", "120"))
|
| 31 |
EAGER_LOAD = os.getenv("SMOLNALYSIS_MINICPM_TRANSFORMERS_EAGER_LOAD", os.getenv("SPACE_ID", "")).casefold() not in {
|
| 32 |
"",
|
|
@@ -133,7 +137,11 @@ def role_config(role: str) -> TransformersRoleConfig:
|
|
| 133 |
|
| 134 |
default_temperature = "0" if role == "ckan_retrieval" else str(DEFAULT_TEMPERATURE)
|
| 135 |
adapter_path = _clean_env_value(_role_env(role, "ADAPTER_PATH"), _clean_env_value(_role_env(role, "LORA_PATH"), ""))
|
| 136 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 137 |
max_new_tokens = int(_clean_env_value(_role_env(role, "MAX_NEW_TOKENS"), str(DEFAULT_MAX_NEW_TOKENS)))
|
| 138 |
temperature = float(_clean_env_value(_role_env(role, "TEMPERATURE"), default_temperature))
|
| 139 |
top_p = float(_clean_env_value(_role_env(role, "TOP_P"), str(DEFAULT_TOP_P)))
|
|
|
|
| 27 |
DEFAULT_MAX_NEW_TOKENS = int(os.getenv("SMOLNALYSIS_MINICPM_MAX_NEW_TOKENS", os.getenv("MAX_TOKENS", "384")))
|
| 28 |
DEFAULT_TEMPERATURE = float(os.getenv("SMOLNALYSIS_MINICPM_TEMPERATURE", os.getenv("TEMPERATURE", "0.7")))
|
| 29 |
DEFAULT_TOP_P = float(os.getenv("SMOLNALYSIS_MINICPM_TOP_P", os.getenv("TOP_P", "0.95")))
|
| 30 |
+
DEFAULT_CKAN_RETRIEVAL_ADAPTER_REPO_ID = os.getenv(
|
| 31 |
+
"SMOLNALYSIS_DEFAULT_CKAN_RETRIEVAL_ADAPTER_REPO_ID",
|
| 32 |
+
"build-small-hackathon/smolnalysis-ckan-retrieval-minicpm5-lora",
|
| 33 |
+
)
|
| 34 |
ZERO_GPU_DURATION_SECONDS = int(os.getenv("SMOLNALYSIS_MINICPM_ZEROGPU_DURATION_SECONDS", "120"))
|
| 35 |
EAGER_LOAD = os.getenv("SMOLNALYSIS_MINICPM_TRANSFORMERS_EAGER_LOAD", os.getenv("SPACE_ID", "")).casefold() not in {
|
| 36 |
"",
|
|
|
|
| 137 |
|
| 138 |
default_temperature = "0" if role == "ckan_retrieval" else str(DEFAULT_TEMPERATURE)
|
| 139 |
adapter_path = _clean_env_value(_role_env(role, "ADAPTER_PATH"), _clean_env_value(_role_env(role, "LORA_PATH"), ""))
|
| 140 |
+
default_adapter_repo_id = DEFAULT_CKAN_RETRIEVAL_ADAPTER_REPO_ID if role == "ckan_retrieval" else ""
|
| 141 |
+
adapter_repo_id = _clean_env_value(
|
| 142 |
+
_role_env(role, "ADAPTER_REPO_ID"),
|
| 143 |
+
_clean_env_value(_role_env(role, "LORA_REPO_ID"), default_adapter_repo_id),
|
| 144 |
+
)
|
| 145 |
max_new_tokens = int(_clean_env_value(_role_env(role, "MAX_NEW_TOKENS"), str(DEFAULT_MAX_NEW_TOKENS)))
|
| 146 |
temperature = float(_clean_env_value(_role_env(role, "TEMPERATURE"), default_temperature))
|
| 147 |
top_p = float(_clean_env_value(_role_env(role, "TOP_P"), str(DEFAULT_TOP_P)))
|