Spaces:
Runtime error
Runtime error
Jarvis-K
commited on
Commit
•
5f98914
1
Parent(s):
d731338
add support to multi-dim con
Browse files- deciders/act.py +16 -8
- deciders/parser.py +35 -12
- envs/base_env.py +5 -3
- main_reflexion.py +4 -8
- record_reflexion.csv +1 -0
deciders/act.py
CHANGED
@@ -3,7 +3,7 @@
|
|
3 |
import openai
|
4 |
from .gpt import gpt
|
5 |
from loguru import logger
|
6 |
-
from .parser import
|
7 |
from langchain.output_parsers import PydanticOutputParser
|
8 |
from langchain.output_parsers import OutputFixingParser
|
9 |
from langchain.chat_models import AzureChatOpenAI, ChatOpenAI
|
@@ -12,13 +12,18 @@ import tiktoken
|
|
12 |
import json
|
13 |
import re
|
14 |
from .utils import run_chain
|
|
|
15 |
|
16 |
class RandomAct():
|
17 |
def __init__(self, action_space):
|
18 |
self.action_space = action_space
|
19 |
|
20 |
def act(self, state_description, action_description, env_info, game_description=None, goal_description=None):
|
21 |
-
|
|
|
|
|
|
|
|
|
22 |
|
23 |
class NaiveAct(gpt):
|
24 |
def __init__(self, action_space, args, prompts, distiller, temperature=0.0, max_tokens=2048, logger=None):
|
@@ -37,7 +42,10 @@ class NaiveAct(gpt):
|
|
37 |
super().__init__(args)
|
38 |
self.distiller = distiller
|
39 |
self.fewshot_example_initialization(args.prompt_level, args.prompt_path, distiller = self.distiller)
|
40 |
-
self.
|
|
|
|
|
|
|
41 |
self.parser = self._parser_initialization()
|
42 |
self.irr_game_description = ''
|
43 |
self.memory = []
|
@@ -82,11 +90,12 @@ class NaiveAct(gpt):
|
|
82 |
|
83 |
|
84 |
def _parser_initialization(self):
|
85 |
-
if
|
86 |
-
|
87 |
num_action = self.action_space.n
|
88 |
-
else:
|
89 |
-
|
|
|
90 |
|
91 |
if self.args.api_type == "azure":
|
92 |
autofixing_chat = AzureChatOpenAI(
|
@@ -204,7 +213,6 @@ class NaiveAct(gpt):
|
|
204 |
prompt, res = self.response(state_description, action_description, env_info, game_description, goal_description, my_mem)
|
205 |
action_str = res.choices[0].text.strip()
|
206 |
print(f'my anwser is {action_str}')
|
207 |
-
# import pdb; pdb.set_trace()
|
208 |
try:
|
209 |
if "Continuous" in self.args.env_name:
|
210 |
action = float(re.findall(r"[-+]?\d*\.\d+", action_str)[0])
|
|
|
3 |
import openai
|
4 |
from .gpt import gpt
|
5 |
from loguru import logger
|
6 |
+
from .parser import DISPARSERS, CONPARSERS
|
7 |
from langchain.output_parsers import PydanticOutputParser
|
8 |
from langchain.output_parsers import OutputFixingParser
|
9 |
from langchain.chat_models import AzureChatOpenAI, ChatOpenAI
|
|
|
12 |
import json
|
13 |
import re
|
14 |
from .utils import run_chain
|
15 |
+
from gym.spaces import Discrete
|
16 |
|
17 |
class RandomAct():
|
18 |
def __init__(self, action_space):
|
19 |
self.action_space = action_space
|
20 |
|
21 |
def act(self, state_description, action_description, env_info, game_description=None, goal_description=None):
|
22 |
+
if isinstance(self.action_space, Discrete):
|
23 |
+
action = self.action_space.sample()+1
|
24 |
+
else:
|
25 |
+
action = self.action_space.sample()
|
26 |
+
return action, '', '', '', 0, 0
|
27 |
|
28 |
class NaiveAct(gpt):
|
29 |
def __init__(self, action_space, args, prompts, distiller, temperature=0.0, max_tokens=2048, logger=None):
|
|
|
42 |
super().__init__(args)
|
43 |
self.distiller = distiller
|
44 |
self.fewshot_example_initialization(args.prompt_level, args.prompt_path, distiller = self.distiller)
|
45 |
+
if isinstance(self.action_space, Discrete):
|
46 |
+
self.default_action = 1
|
47 |
+
else:
|
48 |
+
self.default_action = [0 for ind in range(self.action_space.shape[0])]
|
49 |
self.parser = self._parser_initialization()
|
50 |
self.irr_game_description = ''
|
51 |
self.memory = []
|
|
|
90 |
|
91 |
|
92 |
def _parser_initialization(self):
|
93 |
+
if isinstance(self.action_space, Discrete):
|
94 |
+
PARSERS = DISPARSERS
|
95 |
num_action = self.action_space.n
|
96 |
+
else:
|
97 |
+
PARSERS = CONPARSERS
|
98 |
+
num_action = self.action_space.shape[0]
|
99 |
|
100 |
if self.args.api_type == "azure":
|
101 |
autofixing_chat = AzureChatOpenAI(
|
|
|
213 |
prompt, res = self.response(state_description, action_description, env_info, game_description, goal_description, my_mem)
|
214 |
action_str = res.choices[0].text.strip()
|
215 |
print(f'my anwser is {action_str}')
|
|
|
216 |
try:
|
217 |
if "Continuous" in self.args.env_name:
|
218 |
action = float(re.findall(r"[-+]?\d*\.\d+", action_str)[0])
|
deciders/parser.py
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
from pydantic import BaseModel, Field, validator
|
|
|
2 |
|
3 |
class DisActionModel(BaseModel):
|
4 |
action: int = Field(description="the chosen action to perform")
|
@@ -17,15 +18,37 @@ def generate_action_class(max_action):
|
|
17 |
return type(f"{max_action}Action", (DisActionModel,), {'action_is_valid': DisActionModel.create_validator(max_action)})
|
18 |
|
19 |
# Dictionary of parsers with dynamic class generation
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
from pydantic import BaseModel, Field, validator
|
2 |
+
from typing import List
|
3 |
|
4 |
class DisActionModel(BaseModel):
|
5 |
action: int = Field(description="the chosen action to perform")
|
|
|
18 |
return type(f"{max_action}Action", (DisActionModel,), {'action_is_valid': DisActionModel.create_validator(max_action)})
|
19 |
|
20 |
# Dictionary of parsers with dynamic class generation
|
21 |
+
DISPARSERS = {num: generate_action_class(num) for num in [2, 3, 4, 6, 9, 18]}
|
22 |
+
|
23 |
+
class ContinuousActionBase(BaseModel):
|
24 |
+
action: List[float] = Field(description="the chosen continuous actions to perform")
|
25 |
+
|
26 |
+
@classmethod
|
27 |
+
def set_expected_length(cls, length):
|
28 |
+
cls.expected_length = length
|
29 |
+
|
30 |
+
@validator('action', pre=True)
|
31 |
+
def validate_length(cls, action):
|
32 |
+
if len(action) != cls.expected_length:
|
33 |
+
raise ValueError(f"The action list must have exactly {cls.expected_length} items.")
|
34 |
+
return action
|
35 |
+
|
36 |
+
@validator('action', each_item=True)
|
37 |
+
def action_is_valid(cls, item):
|
38 |
+
if not -1 <= item <= 1:
|
39 |
+
raise ValueError("Each action dimension must be in the range [-1, 1]!")
|
40 |
+
return item
|
41 |
+
|
42 |
+
# Generate classes dynamically
|
43 |
+
def generate_continuous_action_class(expected_length):
|
44 |
+
NewClass = type(
|
45 |
+
f"{expected_length}DContinuousAction",
|
46 |
+
(ContinuousActionBase,),
|
47 |
+
{}
|
48 |
+
)
|
49 |
+
NewClass.set_expected_length(expected_length)
|
50 |
+
return NewClass
|
51 |
+
|
52 |
+
|
53 |
+
# Dictionary of parsers with dynamic class generation
|
54 |
+
CONPARSERS = {length: generate_continuous_action_class(length) for length in range(1, 17)}
|
envs/base_env.py
CHANGED
@@ -1,6 +1,7 @@
|
|
1 |
# This file contains functions for interacting with the CartPole environment
|
2 |
|
3 |
import gym
|
|
|
4 |
|
5 |
class SettableStateEnv(gym.Wrapper):
|
6 |
def __init__(self, env):
|
@@ -55,10 +56,11 @@ class BaseEnv(gym.Wrapper):
|
|
55 |
|
56 |
def step_llm(self, action):
|
57 |
potential_next_state = self.get_potential_next_state(action)
|
58 |
-
if
|
59 |
-
state, reward, terminated, _, info = super().step(action)
|
60 |
-
else:
|
61 |
state, reward, terminated, _, info = super().step(action-1)
|
|
|
|
|
|
|
62 |
self.transition_data['action'] = action
|
63 |
self.transition_data['next_state'] = state
|
64 |
self.transition_data['reward'] = reward
|
|
|
1 |
# This file contains functions for interacting with the CartPole environment
|
2 |
|
3 |
import gym
|
4 |
+
from gym.spaces import Discrete
|
5 |
|
6 |
class SettableStateEnv(gym.Wrapper):
|
7 |
def __init__(self, env):
|
|
|
56 |
|
57 |
def step_llm(self, action):
|
58 |
potential_next_state = self.get_potential_next_state(action)
|
59 |
+
if isinstance(self.action_space, Discrete):
|
|
|
|
|
60 |
state, reward, terminated, _, info = super().step(action-1)
|
61 |
+
else:
|
62 |
+
state, reward, terminated, _, info = super().step(action)
|
63 |
+
|
64 |
self.transition_data['action'] = action
|
65 |
self.transition_data['next_state'] = state
|
66 |
self.transition_data['reward'] = reward
|
main_reflexion.py
CHANGED
@@ -17,6 +17,9 @@ import random
|
|
17 |
import numpy as np
|
18 |
import datetime
|
19 |
from loguru import logger
|
|
|
|
|
|
|
20 |
|
21 |
|
22 |
def set_seed(seed):
|
@@ -109,9 +112,6 @@ def _run(translator, environment, decider, max_episode_len, logfile, args, trail
|
|
109 |
logfile
|
110 |
)
|
111 |
|
112 |
-
if "Continuous" in args.env_name:
|
113 |
-
action = [action]
|
114 |
-
|
115 |
state_description, reward, termination, truncation, env_info = environment.step_llm(
|
116 |
action
|
117 |
)
|
@@ -137,10 +137,6 @@ def _run(translator, environment, decider, max_episode_len, logfile, args, trail
|
|
137 |
logger.debug(f"Error: {e}, Retry! ({error_i+1}/{retry_num})")
|
138 |
continue
|
139 |
if error_flag:
|
140 |
-
if "Continuous" in args.env_name:
|
141 |
-
action = [decider.default_action]
|
142 |
-
else:
|
143 |
-
action = decider.default_action
|
144 |
state_description, reward, termination, truncation, env_info = environment.step_llm(
|
145 |
action
|
146 |
)
|
@@ -164,7 +160,7 @@ def _run(translator, environment, decider, max_episode_len, logfile, args, trail
|
|
164 |
logger.info(f"current_total_cost: {current_total_cost}")
|
165 |
logger.info(f"Now it is round {round}.")
|
166 |
|
167 |
-
frames.append(environment.render())
|
168 |
if termination or truncation:
|
169 |
if logger:
|
170 |
logger.info(f"Terminated!")
|
|
|
17 |
import numpy as np
|
18 |
import datetime
|
19 |
from loguru import logger
|
20 |
+
from gym.spaces import Discrete
|
21 |
+
|
22 |
+
|
23 |
|
24 |
|
25 |
def set_seed(seed):
|
|
|
112 |
logfile
|
113 |
)
|
114 |
|
|
|
|
|
|
|
115 |
state_description, reward, termination, truncation, env_info = environment.step_llm(
|
116 |
action
|
117 |
)
|
|
|
137 |
logger.debug(f"Error: {e}, Retry! ({error_i+1}/{retry_num})")
|
138 |
continue
|
139 |
if error_flag:
|
|
|
|
|
|
|
|
|
140 |
state_description, reward, termination, truncation, env_info = environment.step_llm(
|
141 |
action
|
142 |
)
|
|
|
160 |
logger.info(f"current_total_cost: {current_total_cost}")
|
161 |
logger.info(f"Now it is round {round}.")
|
162 |
|
163 |
+
# frames.append(environment.render())
|
164 |
if termination or truncation:
|
165 |
if logger:
|
166 |
logger.info(f"Terminated!")
|
record_reflexion.csv
CHANGED
@@ -10,4 +10,5 @@ FrozenLake-v1,1,expert,200.0
|
|
10 |
MountainCarContinuous-v0,1,expert,200.0
|
11 |
RepresentedBoxing-v0,1,expert,200.0
|
12 |
RepresentedPong-v0,1,expert,200.0
|
|
|
13 |
|
|
|
10 |
MountainCarContinuous-v0,1,expert,200.0
|
11 |
RepresentedBoxing-v0,1,expert,200.0
|
12 |
RepresentedPong-v0,1,expert,200.0
|
13 |
+
Ant-v4,1,expert,100
|
14 |
|