Skip to content

armgit / dspy

Public Imported from GitHub
0

DSPy: The framework for programming—not prompting—language models

Python

pre-2.5

Isaac Miller86e7740b8aba
diff --git a/dsp/modules/lm.py b/dsp/modules/lm.py
index 3ab3e41cb..92049eab1 100644
--- a/dsp/modules/lm.py
+++ b/dsp/modules/lm.py
@@ -55,7 +55,7 @@ class LM(ABC):
printed = []
n = n + skip
- for x in reversed(self.history[-100:]):
+ for x in reversed(self.history):
prompt = x["prompt"]
if prompt != last_prompt:
diff --git a/dsp/modules/trainable_anyscale.py b/dsp/modules/trainable_anyscale.py
index 361bb0c51..325ad3f9c 100644
--- a/dsp/modules/trainable_anyscale.py
+++ b/dsp/modules/trainable_anyscale.py
@@ -119,8 +119,14 @@ class TrainableAnyscale(MultiOpenAI, TrainableLM):
custom_modifications = {
"model_id": self.kwargs["model"],
"train_path": train_path,
- "output_dir": kwargs.get("output_dir", "/mnt/cluster_storage/dspy/finetuning/artifacts/")
+ "logger": {
+ "provider": "wandb",
+ },
+ "num_checkpoints_to_keep": 10
+ # "output_dir": kwargs.get("output_dir", "/mnt/cluster_storage/dspy/finetuning/artifacts/")
}
+ if kwargs.get("output_dir", None):
+ custom_modifications["output_dir"] = kwargs["output_dir"]
if eval_path:
custom_modifications["valid_path"] = eval_path
@@ -158,14 +164,22 @@ class TrainableAnyscale(MultiOpenAI, TrainableLM):
"name": "dspy-llmforge-fine-tuning-job",
"entrypoint": f"llmforge anyscale finetune {filename}",
"working_dir": ".",
- "image_uri": "localhost:5555/anyscale/llm-forge:0.5.5"
+ "image_uri": "localhost:5555/anyscale/llm-forge:0.5.6",
+ "requirements": [
+ "wandb",
+ ],
+ "env_vars": {
+ "WANDB_API_KEY": os.environ.get("WANDB_API_KEY", ""),
+ "HF_TOKEN": os.environ.get("HF_TOKEN", ""),
+ "HF_HOME": os.environ.get("HF_HOME", ""),
+ }
}
compute_config_kwargs = kwargs.get("compute_config", {})
compute_config_dict.update(compute_config_kwargs)
compute_config = JobConfig(**compute_config_dict)
job_runner_config_path = kwargs.get("compute_yaml_path", "job_runner_config.yaml")
- yaml.safe_dump(compute_config_dict, open(job_runner_config_path, "w"))
+ # yaml.safe_dump(compute_config_dict, open(job_runner_config_path, "w"))
# TODO: Validate the hyperparameters
# if not self.validate_hyperparameters(training_arguments):
@@ -219,14 +233,14 @@ class TrainableAnyscale(MultiOpenAI, TrainableLM):
print(f"Number of items in {name} data: {num_items}")
- s3_path = os.path.join(storage, f"{name}.jsonl")
+ s3_path = os.path.join(storage, path.split("/")[-1])
print(f"Uploading {name} data to S3 at {s3_path}")
# ray.data.read_json(path).write_json(s3_path)
# NOTE: trying a local copy for now
if s3_path[:2] == "s3":
os.system(f"aws s3 cp {path} {s3_path}")
elif s3_path[:2] == "gs":
- os.system(f"gsutil cp {path} {s3_path}")
+ os.system(f"gcloud storage cp {path} {s3_path}")
else:
os.system(f"cp {path} {s3_path}")
print(f"Copied {path} to {s3_path}")
diff --git a/examples/anyscale/e2e/e2e_dspy.ipynb b/examples/anyscale/e2e/e2e_dspy.ipynb
index 4044f55fa..bcb385e92 100644
--- a/examples/anyscale/e2e/e2e_dspy.ipynb
+++ b/examples/anyscale/e2e/e2e_dspy.ipynb
@@ -9362,7 +9362,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
- "version": "3.9.19"
+ "version": "3.11.9"
}
},
"nbformat": 4,