Compare commits
1 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
| fb1efa3b4a |
@@ -2,5 +2,5 @@
|
|||||||
"model": "openai//models/checkpoint-5",
|
"model": "openai//models/checkpoint-5",
|
||||||
"api_base": "https://modaic-ai--grpo-demo-serve.modal.run/v1",
|
"api_base": "https://modaic-ai--grpo-demo-serve.modal.run/v1",
|
||||||
"max_tokens": 10000,
|
"max_tokens": 10000,
|
||||||
"temperature": 2
|
"temperature": 0.7
|
||||||
}
|
}
|
||||||
4
main.py
4
main.py
@@ -5,7 +5,7 @@ class CodeGeneratorConfig(PrecompiledConfig):
|
|||||||
model : str = "openai//models/checkpoint-5"
|
model : str = "openai//models/checkpoint-5"
|
||||||
api_base : str = "https://modaic-ai--grpo-demo-serve.modal.run/v1"
|
api_base : str = "https://modaic-ai--grpo-demo-serve.modal.run/v1"
|
||||||
max_tokens : int = 10000
|
max_tokens : int = 10000
|
||||||
temperature : float = 2
|
temperature : float = 0.7
|
||||||
|
|
||||||
|
|
||||||
class CodeGeneration(dspy.Signature):
|
class CodeGeneration(dspy.Signature):
|
||||||
@@ -32,6 +32,6 @@ class CodeGenerator(PrecompiledProgram):
|
|||||||
|
|
||||||
code_generator = CodeGenerator(CodeGeneratorConfig())
|
code_generator = CodeGenerator(CodeGeneratorConfig())
|
||||||
print(code_generator(query="Write a python function that returns the sum of two numbers.").code)
|
print(code_generator(query="Write a python function that returns the sum of two numbers.").code)
|
||||||
code_generator.push_to_hub("modaic/code-generator-trl-grpo", with_code=True, tag="v2")
|
code_generator.push_to_hub("modaic/code-generator-trl-grpo", with_code=True, tag="v2.0.1")
|
||||||
|
|
||||||
|
|
||||||
@@ -24,7 +24,7 @@
|
|||||||
"finetuning_model": null,
|
"finetuning_model": null,
|
||||||
"launch_kwargs": {},
|
"launch_kwargs": {},
|
||||||
"train_kwargs": {},
|
"train_kwargs": {},
|
||||||
"temperature": 2,
|
"temperature": 0.7,
|
||||||
"max_tokens": 10000,
|
"max_tokens": 10000,
|
||||||
"api_base": "https://modaic-ai--grpo-demo-serve.modal.run/v1"
|
"api_base": "https://modaic-ai--grpo-demo-serve.modal.run/v1"
|
||||||
}
|
}
|
||||||
|
|||||||
Reference in New Issue
Block a user