-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathplot_bounds.py
383 lines (295 loc) · 18.1 KB
/
plot_bounds.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
import jax.numpy as jnp
import numpy as np
from flax.training import checkpoints
import matplotlib
matplotlib.use('PDF')
import matplotlib.pyplot as plt
from plot_utils import plot_with_conf_bounds
plot_type = "toxthresh" # "sent_dpg_comparison" # "sent" #"toxclass" #
if plot_type == "toxclass":
# TOX CLASS
load_pref_twist_1_8 = "logZ_bounds_twistproposal_2024-03-26_15-33_seed1_ebm_one_sample_nsamples1"
load_pref_p_1_8 = "logZ_bounds_pproposal_2024-03-26_15-37_seed1_ebm_one_sample_nsamples1"
load_pref_twist_4_16 = "logZ_bounds_twistproposal_2024-03-26_15-33_seed1_ebm_one_sample_nsamples1"
load_pref_p_4_16 = "logZ_bounds_pproposal_2024-03-26_15-38_seed1_ebm_one_sample_nsamples1"
figname = f"./fig_bounds_toxclass_03-26.pdf"
elif plot_type == "sent":
load_pref_twist_1_8 = "logZ_bounds_twistproposal_2024-03-26_16-00_seed1_ebm_one_sample_nsamples1"
load_pref_p_1_8 = "logZ_bounds_pproposal_2024-03-26_22-12_seed1_ebm_one_sample_nsamples1"
load_pref_twist_4_16 = "logZ_bounds_twistproposal_2024-03-26_15-53_seed1_ebm_one_sample_nsamples1"
load_pref_p_4_16 = "logZ_bounds_pproposal_2024-03-26_22-10_seed1_ebm_one_sample_nsamples1"
figname = f"./fig_bounds_sent_03-26.pdf"
elif plot_type == "sent_dpg_comparison":
load_pref_twist_1_8 = "logZ_bounds_twistproposal_2024-03-26_16-00_seed1_ebm_one_sample_nsamples1"
load_pref_p_1_8 = "logZ_bounds_pproposal_2024-03-26_22-12_seed1_ebm_one_sample_nsamples1"
load_pref_twist_4_16 = "logZ_bounds_twistproposal_2024-03-26_15-53_seed1_ebm_one_sample_nsamples1"
load_pref_p_4_16 = "logZ_bounds_pproposal_2024-03-26_22-10_seed1_ebm_one_sample_nsamples1"
load_pref_twist_1_8_dpg = "logZ_bounds_twistproposal_2024-03-28_00-03_seed1_dpg_nsamples1"
load_pref_p_1_8_dpg = "logZ_bounds_pproposal_2024-03-28_00-03_seed1_dpg_nsamples1_8"
load_pref_twist_4_16_dpg = "logZ_bounds_twistproposal_2024-03-28_00-02_seed1_dpg_nsamples1"
load_pref_p_4_16_dpg = "logZ_bounds_pproposal_2024-03-28_00-03_seed1_dpg_nsamples4_16"
figname = f"./fig_bounds_sent_dpg_comparison_03-27.pdf"
elif plot_type == "toxthresh":
# TOX THRESH
load_pref_twist_1_8 = "logZ_bounds_twistproposal_2024-03-24_18-44_seed1_ebm_one_sample_nsamples1"
load_pref_p_1_8 = "logZ_bounds_pproposal_2024-03-24_18-48_seed1_ebm_one_sample_nsamples1"
load_pref_twist_4_16 = "logZ_bounds_twistproposal_2024-03-25_14-23_seed1_ebm_one_sample_nsamples1"
load_pref_p_4_16 = "logZ_bounds_pproposal_2024-03-25_14-27_seed1_ebm_one_sample_nsamples1"
load_pref_twist_32_512 = "logZ_bounds_twistproposal_2024-01-15_12-52_seed1_ebm_one_sample_nsamples1_1"
load_pref_p_32_512 = "logZ_bounds_pproposal_2024-01-15_12-58_seed1_ebm_one_sample_nsamples1_1"
load_pref_twist_128_2048 = "logZ_bounds_twistproposal_2024-01-15_12-59_seed1_ebm_one_sample_nsamples1_2"
load_pref_p_128_2048 = "logZ_bounds_pproposal_2024-01-15_13-10_seed1_ebm_one_sample_nsamples1_2"
load_pref_p_1_8_ess = "logZ_bounds_pproposal_2024-03-25_15-00_seed1_ebm_one_sample_nsamples1"
load_pref_p_4_16_ess = "logZ_bounds_pproposal_2024-03-25_14-59_seed1_ebm_one_sample_nsamples1"
load_pref_p_32_512_ess = "logZ_bounds_pproposal_2024-02-12_23-43_seed1_ebm_one_sample_nsamples1_1"
load_pref_p_128_2048_ess = "logZ_bounds_pproposal_2024-02-12_23-52_seed1_ebm_one_sample_nsamples1_2"
plot_ess = True
figname = f"./fig_bounds_with_ess_toxt_-5_04-03.pdf"
# plot_ess = False
# figname = f"./fig_bounds_no_ess_toxt_-5_04-03.pdf"
# figname = f"./fig_bounds_toxt_-5_03-24.pdf"
color_list_for_lbs = ['xkcd:light blue', 'xkcd:light green', 'xkcd:light orange', 'xkcd:light red', 'xkcd:light purple', 'xkcd:dark grey']
color_list_for_ubs = ['xkcd:blue', 'xkcd:green', 'xkcd:orange', 'xkcd:red', 'xkcd:purple', 'xkcd:black']
linestyle_list_for_lbs = ['solid'] * 10
linestyle_list_for_ubs = ['dashed'] * 10
load_dir = "./f_q_g_q_logZ_info"
def load_ckpt(load_prefix):
x = checkpoints.restore_checkpoint(ckpt_dir=f"{load_dir}/{load_prefix}", target=None,
prefix="checkpoint")
logZ_ubs_iwae_across_samples_time_seeds, logZ_lbs_iwae_across_samples_time_seeds, \
logZ_ubs_smc_across_samples_time_seeds, logZ_lbs_smc_across_samples_time_seeds = x
logZ_ubs_iwae_across_samples_time_seeds_smaller, logZ_ubs_iwae_across_samples_time_seeds_larger = logZ_ubs_iwae_across_samples_time_seeds
logZ_lbs_iwae_across_samples_time_seeds_smaller, logZ_lbs_iwae_across_samples_time_seeds_larger = logZ_lbs_iwae_across_samples_time_seeds
logZ_ubs_smc_across_samples_time_seeds_smaller, logZ_ubs_smc_across_samples_time_seeds_larger = logZ_ubs_smc_across_samples_time_seeds
logZ_lbs_smc_across_samples_time_seeds_smaller, logZ_lbs_smc_across_samples_time_seeds_larger = logZ_lbs_smc_across_samples_time_seeds
return logZ_ubs_iwae_across_samples_time_seeds_smaller[0], logZ_ubs_iwae_across_samples_time_seeds_larger[0], \
logZ_lbs_iwae_across_samples_time_seeds_smaller[0], logZ_lbs_iwae_across_samples_time_seeds_larger[0], \
logZ_ubs_smc_across_samples_time_seeds_smaller[0], logZ_ubs_smc_across_samples_time_seeds_larger[0], \
logZ_lbs_smc_across_samples_time_seeds_smaller[0], logZ_lbs_smc_across_samples_time_seeds_larger[0]
t_iwae_ubs_1, t_iwae_ubs_8, t_iwae_lbs_1, t_iwae_lbs_8, t_smc_ubs_1, t_smc_ubs_8, t_smc_lbs_1, t_smc_lbs_8 = load_ckpt(load_pref_twist_1_8)
p_iwae_ubs_1, p_iwae_ubs_8, p_iwae_lbs_1, p_iwae_lbs_8, p_smc_ubs_1, p_smc_ubs_8, p_smc_lbs_1, p_smc_lbs_8 = load_ckpt(load_pref_p_1_8)
t_iwae_ubs_4, t_iwae_ubs_16, t_iwae_lbs_4, t_iwae_lbs_16, t_smc_ubs_4, t_smc_ubs_16, t_smc_lbs_4, t_smc_lbs_16 = load_ckpt(load_pref_twist_4_16)
p_iwae_ubs_4, p_iwae_ubs_16, p_iwae_lbs_4, p_iwae_lbs_16, p_smc_ubs_4, p_smc_ubs_16, p_smc_lbs_4, p_smc_lbs_16 = load_ckpt(load_pref_p_4_16)
if plot_type in ["toxclass", "sent"]:
x_range = np.array([0,2,3,4])
t_iwae_ubs = np.transpose(np.stack([t_iwae_ubs_1, t_iwae_ubs_4, t_iwae_ubs_8, t_iwae_ubs_16]))
t_iwae_lbs = np.transpose(np.stack([t_iwae_lbs_1, t_iwae_lbs_4, t_iwae_lbs_8, t_iwae_lbs_16]))
t_smc_ubs = np.transpose(np.stack([t_smc_ubs_1, t_smc_ubs_4, t_smc_ubs_8, t_smc_ubs_16]))
t_smc_lbs = np.transpose(np.stack([t_smc_lbs_1, t_smc_lbs_4, t_smc_lbs_8, t_smc_lbs_16]))
p_iwae_ubs = np.transpose(np.stack([p_iwae_ubs_1, p_iwae_ubs_4, p_iwae_ubs_8, p_iwae_ubs_16]))
p_iwae_lbs = np.transpose(np.stack([p_iwae_lbs_1, p_iwae_lbs_4, p_iwae_lbs_8, p_iwae_lbs_16]))
p_smc_ubs = np.transpose(np.stack([p_smc_ubs_1, p_smc_ubs_4, p_smc_ubs_8, p_smc_ubs_16]))
p_smc_lbs = np.transpose(np.stack([p_smc_lbs_1, p_smc_lbs_4, p_smc_lbs_8, p_smc_lbs_16]))
# print("---")
# for x in [t_iwae_ubs_1, t_iwae_ubs_8, t_iwae_lbs_1, t_iwae_lbs_8, t_smc_ubs_1, t_smc_ubs_8, t_smc_lbs_1, t_smc_lbs_8]:
# print(x)
# print(jnp.stack(x).mean())
# print("---")
# for x in [p_iwae_ubs_1, p_iwae_ubs_8, p_iwae_lbs_1, p_iwae_lbs_8, p_smc_ubs_1, p_smc_ubs_8, p_smc_lbs_1, p_smc_lbs_8]:
# print(x)
# print(jnp.stack(x).mean())
# 1/0
elif plot_type == "toxthresh":
x_range = np.array([0, 2, 3, 4, 5, 7, 9, 11])
t_iwae_ubs_32, t_iwae_ubs_512, t_iwae_lbs_32, t_iwae_lbs_512, t_smc_ubs_32, t_smc_ubs_512, t_smc_lbs_32, t_smc_lbs_512 = load_ckpt(load_pref_twist_32_512)
p_iwae_ubs_32, p_iwae_ubs_512, p_iwae_lbs_32, p_iwae_lbs_512, p_smc_ubs_32, p_smc_ubs_512, p_smc_lbs_32, p_smc_lbs_512 = load_ckpt(load_pref_p_32_512)
t_iwae_ubs_128, t_iwae_ubs_2048, t_iwae_lbs_128, t_iwae_lbs_2048, t_smc_ubs_128, t_smc_ubs_2048, t_smc_lbs_128, t_smc_lbs_2048 = load_ckpt(load_pref_twist_128_2048)
p_iwae_ubs_128, p_iwae_ubs_2048, p_iwae_lbs_128, p_iwae_lbs_2048, p_smc_ubs_128, p_smc_ubs_2048, p_smc_lbs_128, p_smc_lbs_2048 = load_ckpt(load_pref_p_128_2048)
t_iwae_ubs = np.transpose(np.stack([t_iwae_ubs_1, t_iwae_ubs_4, t_iwae_ubs_8, t_iwae_ubs_16, t_iwae_ubs_32, t_iwae_ubs_128, t_iwae_ubs_512, t_iwae_ubs_2048]))
t_iwae_lbs = np.transpose(np.stack([t_iwae_lbs_1, t_iwae_lbs_4, t_iwae_lbs_8, t_iwae_lbs_16, t_iwae_lbs_32, t_iwae_lbs_128, t_iwae_lbs_512, t_iwae_lbs_2048]))
t_smc_ubs = np.transpose(np.stack([t_smc_ubs_1, t_smc_ubs_4, t_smc_ubs_8, t_smc_ubs_16, t_smc_ubs_32, t_smc_ubs_128, t_smc_ubs_512, t_smc_ubs_2048]))
t_smc_lbs = np.transpose(np.stack([t_smc_lbs_1, t_smc_lbs_4, t_smc_lbs_8, t_smc_lbs_16, t_smc_lbs_32, t_smc_lbs_128, t_smc_lbs_512, t_smc_lbs_2048]))
p_iwae_ubs = np.transpose(np.stack([p_iwae_ubs_1, p_iwae_ubs_4, p_iwae_ubs_8, p_iwae_ubs_16, p_iwae_ubs_32, p_iwae_ubs_128, p_iwae_ubs_512, p_iwae_ubs_2048]))
p_iwae_lbs = np.transpose(np.stack([p_iwae_lbs_1, p_iwae_lbs_4, p_iwae_lbs_8, p_iwae_lbs_16, p_iwae_lbs_32, p_iwae_lbs_128, p_iwae_lbs_512, p_iwae_lbs_2048]))
p_smc_ubs = np.transpose(np.stack([p_smc_ubs_1, p_smc_ubs_4, p_smc_ubs_8, p_smc_ubs_16, p_smc_ubs_32, p_smc_ubs_128, p_smc_ubs_512, p_smc_ubs_2048]))
p_smc_lbs = np.transpose(np.stack([p_smc_lbs_1, p_smc_lbs_4, p_smc_lbs_8, p_smc_lbs_16, p_smc_lbs_32, p_smc_lbs_128, p_smc_lbs_512, p_smc_lbs_2048]))
p_iwae_ubs_1_ess, p_iwae_ubs_8_ess, p_iwae_lbs_1_ess, p_iwae_lbs_8_ess, p_smc_ubs_1_ess, p_smc_ubs_8_ess, p_smc_lbs_1_ess, p_smc_lbs_8_ess = load_ckpt(
load_pref_p_1_8_ess)
p_iwae_ubs_4_ess, p_iwae_ubs_16_ess, p_iwae_lbs_4_ess, p_iwae_lbs_16_ess, p_smc_ubs_4_ess, p_smc_ubs_16_ess, p_smc_lbs_4_ess, p_smc_lbs_16_ess = load_ckpt(
load_pref_p_4_16_ess)
p_iwae_ubs_32_ess, p_iwae_ubs_512_ess, p_iwae_lbs_32_ess, p_iwae_lbs_512_ess, p_smc_ubs_32_ess, p_smc_ubs_512_ess, p_smc_lbs_32_ess, p_smc_lbs_512_ess = load_ckpt(
load_pref_p_32_512_ess)
p_iwae_ubs_128_ess, p_iwae_ubs_2048_ess, p_iwae_lbs_128_ess, p_iwae_lbs_2048_ess, p_smc_ubs_128_ess, p_smc_ubs_2048_ess, p_smc_lbs_128_ess, p_smc_lbs_2048_ess = load_ckpt(
load_pref_p_128_2048_ess)
p_smc_ubs_ess = np.transpose(np.stack([p_smc_ubs_1_ess, p_smc_ubs_4_ess, p_smc_ubs_8_ess, p_smc_ubs_16_ess, p_smc_ubs_32_ess, p_smc_ubs_128_ess, p_smc_ubs_512_ess, p_smc_ubs_2048_ess]))
p_smc_lbs_ess = np.transpose(np.stack([p_smc_lbs_1_ess, p_smc_lbs_4_ess, p_smc_lbs_8_ess, p_smc_lbs_16_ess, p_smc_lbs_32_ess, p_smc_lbs_128_ess, p_smc_lbs_512_ess, p_smc_lbs_2048_ess]))
elif plot_type == "sent_dpg_comparison":
t_iwae_ubs_1_dpg, t_iwae_ubs_8_dpg, t_iwae_lbs_1_dpg, t_iwae_lbs_8_dpg, t_smc_ubs_1_dpg, t_smc_ubs_8_dpg, t_smc_lbs_1_dpg, t_smc_lbs_8_dpg = load_ckpt(
load_pref_twist_1_8_dpg)
p_iwae_ubs_1_dpg, p_iwae_ubs_8_dpg, p_iwae_lbs_1_dpg, p_iwae_lbs_8_dpg, p_smc_ubs_1_dpg, p_smc_ubs_8_dpg, p_smc_lbs_1_dpg, p_smc_lbs_8_dpg = load_ckpt(
load_pref_p_1_8_dpg)
t_iwae_ubs_4_dpg, t_iwae_ubs_16_dpg, t_iwae_lbs_4_dpg, t_iwae_lbs_16_dpg, t_smc_ubs_4_dpg, t_smc_ubs_16_dpg, t_smc_lbs_4_dpg, t_smc_lbs_16_dpg = load_ckpt(
load_pref_twist_4_16)
p_iwae_ubs_4_dpg, p_iwae_ubs_16_dpg, p_iwae_lbs_4_dpg, p_iwae_lbs_16_dpg, p_smc_ubs_4_dpg, p_smc_ubs_16_dpg, p_smc_lbs_4_dpg, p_smc_lbs_16_dpg = load_ckpt(
load_pref_p_4_16)
x_range = np.array([0,2,3,4])
t_iwae_ubs = np.transpose(np.stack([t_iwae_ubs_1, t_iwae_ubs_4, t_iwae_ubs_8, t_iwae_ubs_16]))
t_iwae_lbs = np.transpose(np.stack([t_iwae_lbs_1, t_iwae_lbs_4, t_iwae_lbs_8, t_iwae_lbs_16]))
p_iwae_ubs = np.transpose(np.stack([p_iwae_ubs_1, p_iwae_ubs_4, p_iwae_ubs_8, p_iwae_ubs_16]))
p_iwae_lbs = np.transpose(np.stack([p_iwae_lbs_1, p_iwae_lbs_4, p_iwae_lbs_8, p_iwae_lbs_16]))
t_iwae_ubs_dpg = np.transpose(np.stack([t_iwae_ubs_1_dpg, t_iwae_ubs_4_dpg, t_iwae_ubs_8_dpg, t_iwae_ubs_16_dpg]))
t_iwae_lbs_dpg = np.transpose(np.stack([t_iwae_lbs_1_dpg, t_iwae_lbs_4_dpg, t_iwae_lbs_8_dpg, t_iwae_lbs_16_dpg]))
p_iwae_ubs_dpg = np.transpose(np.stack([p_iwae_ubs_1_dpg, p_iwae_ubs_4_dpg, p_iwae_ubs_8_dpg, p_iwae_ubs_16_dpg]))
p_iwae_lbs_dpg = np.transpose(np.stack([p_iwae_lbs_1_dpg, p_iwae_lbs_4_dpg, p_iwae_lbs_8_dpg, p_iwae_lbs_16_dpg]))
plt.clf()
plt.xlabel(f"Number of Samples")
xticks_range = x_range
xticks_labels = 2 ** xticks_range
plt.ylabel(f"Log Z Bound")
if plot_type == "sent_dpg_comparison":
last, conf_bound = plot_with_conf_bounds(
t_iwae_ubs, x_range, label=f"CTL SIS/IWAE UB ($q^\pi$ Proposal)",
color=color_list_for_ubs[0],
linestyle=linestyle_list_for_ubs[0],
)
last, conf_bound = plot_with_conf_bounds(
t_iwae_lbs, x_range, label=f"CTL SIS/IWAE LB ($q^\pi$ Proposal)",
color=color_list_for_lbs[0],
linestyle=linestyle_list_for_lbs[0],
)
last, conf_bound = plot_with_conf_bounds(
p_iwae_ubs, x_range, label=f"CTL SIS/IWAE UB (Base $p_0$ Proposal)",
color=color_list_for_ubs[2],
linestyle=linestyle_list_for_ubs[2],
)
last, conf_bound = plot_with_conf_bounds(
p_iwae_lbs, x_range, label=f"CTL SIS/IWAE LB (Base $p_0$ Proposal)",
color=color_list_for_lbs[2],
linestyle=linestyle_list_for_lbs[2],
)
last, conf_bound = plot_with_conf_bounds(
t_iwae_ubs_dpg, x_range, label=f"DPG SIS/IWAE UB ($q^\pi$ Proposal)",
color=color_list_for_ubs[1],
linestyle=linestyle_list_for_ubs[1],
)
last, conf_bound = plot_with_conf_bounds(
t_iwae_lbs_dpg, x_range, label=f"DPG SIS/IWAE LB ($q^\pi$ Proposal)",
color=color_list_for_lbs[1],
linestyle=linestyle_list_for_lbs[1],
)
last, conf_bound = plot_with_conf_bounds(
p_iwae_ubs_dpg, x_range, label=f"DPG SIS/IWAE UB (Base $p_0$ Proposal)",
color=color_list_for_ubs[3],
linestyle=linestyle_list_for_ubs[3],
)
last, conf_bound = plot_with_conf_bounds(
p_iwae_lbs_dpg, x_range, label=f"DPG SIS/IWAE LB (Base $p_0$ Proposal)",
color=color_list_for_lbs[3],
linestyle=linestyle_list_for_lbs[3],
)
else:
start_from = 1
x_range = x_range[start_from:]
xticks_range = xticks_range[start_from:]
xticks_labels = xticks_labels[start_from:]
print(xticks_labels)
print(xticks_labels.shape)
print(x_range)
print(x_range.shape)
# 1 / 0
last, conf_bound = plot_with_conf_bounds(
t_iwae_ubs[:, start_from:], x_range, label=f"SIS/IWAE UB ($q^\pi$ Proposal)",
color=color_list_for_ubs[0],
linestyle=linestyle_list_for_ubs[0],
)
last, conf_bound = plot_with_conf_bounds(
t_iwae_lbs[:, start_from:], x_range, label=f"SIS/IWAE LB ($q^\pi$ Proposal)",
color=color_list_for_lbs[0],
linestyle=linestyle_list_for_lbs[0],
)
last, conf_bound = plot_with_conf_bounds(
t_smc_ubs[:, start_from:], x_range, label=f"SMC UB ($q^\pi$ Proposal)",
color=color_list_for_ubs[1],
linestyle=linestyle_list_for_ubs[1],
)
last, conf_bound = plot_with_conf_bounds(
t_smc_lbs[:, start_from:], x_range, label=f"SMC LB ($q^\pi$ Proposal)",
color=color_list_for_lbs[1],
linestyle=linestyle_list_for_lbs[1],
)
last, conf_bound = plot_with_conf_bounds(
p_iwae_ubs[:, start_from:], x_range, label=f"SIS/IWAE UB ($p_0$ Proposal)",
color=color_list_for_ubs[2],
linestyle=linestyle_list_for_ubs[2],
)
last, conf_bound = plot_with_conf_bounds(
p_iwae_lbs[:, start_from:], x_range, label=f"SIS/IWAE LB ($p_0$ Proposal)",
color=color_list_for_lbs[2],
linestyle=linestyle_list_for_lbs[2],
)
last, conf_bound = plot_with_conf_bounds(
p_smc_ubs[:, start_from:], x_range, label=f"SMC UB ($p_0$ Proposal)",
color=color_list_for_ubs[3],
linestyle=linestyle_list_for_ubs[3],
)
last, conf_bound = plot_with_conf_bounds(
p_smc_lbs[:, start_from:], x_range, label=f"SMC LB ($p_0$ Proposal)",
color=color_list_for_lbs[3],
linestyle=linestyle_list_for_lbs[3],
)
if plot_type == "toxthresh":
plt.ylim([-36, 1])
if plot_ess:
last, conf_bound = plot_with_conf_bounds(
p_smc_ubs_ess[:, start_from:], x_range, label=f"SMC ESS UB ($p_0$ Proposal)",
color=color_list_for_ubs[4],
linestyle=linestyle_list_for_ubs[4],
)
last, conf_bound = plot_with_conf_bounds(
p_smc_lbs_ess[:, start_from:], x_range, label=f"SMC ESS LB ($p_0$ Proposal)",
color=color_list_for_lbs[4],
linestyle=linestyle_list_for_lbs[4],
)
plt.xticks(xticks_range, xticks_labels)
plt.legend(loc='upper right', bbox_to_anchor=(1, 0.6), fontsize=7)
plt.savefig(figname)
#
# twist_learn_method_name = twist_learn_method_names[i]
#
# last_avg_kl_q_sigma, conf_bound_q_sigma = plot_with_conf_bounds(
# logZ_midpoint_estimate - f_q_estimates, x_range, label=f"{twist_learn_method_name} KL(q||sigma)", # Best logZ meaning using the midpoint of the tightest LogZ bounds that we had.
# color=color_list_for_f_q[i],
# linestyle=linestyle_list_for_f_q[i]
# )
# last_avg_kl_sigma_q, conf_bound_sigma_q = plot_with_conf_bounds(
# g_q_estimates - logZ_midpoint_estimate, x_range, label=f"{twist_learn_method_name} KL(sigma||q)",
# color=color_list_for_g_q[i],
# linestyle=linestyle_list_for_g_q[i]
# )
#
# output_latex.append(f"{twist_learn_method_name} & ${last_avg_kl_q_sigma:.2f} \pm {conf_bound_q_sigma:.2f}$ & ${last_avg_kl_sigma_q:.2f} \pm {conf_bound_sigma_q:.2f}$ \\\\ \midrule")
#
# plt.legend()
# plt.savefig(f"./fig_kl_{fig_name_modifier}_{f_q_estimates.shape[-1]}.pdf")
#
# plt.clf()
# plt.xlabel(f"2^ of Number of Twist Updates")
# plt.ylabel(f"Average Reward")
#
# for i in range(len(load_prefixes)):
#
# reward = reward_list[i]
#
# x_range = np.arange(reward.shape[-1])
#
# twist_learn_method_name = twist_learn_method_names[i]
#
# plot_with_conf_bounds(
# reward, x_range, label=f"{twist_learn_method_name}", # Best logZ meaning using the midpoint of the tightest LogZ bounds that we had.
# color=color_list_for_g_q[i],
# linestyle=linestyle_list_for_f_q[i]
# )
#
# plt.legend()
# plt.savefig(f"./fig_rew_{fig_name_modifier}_{reward.shape[-1]}.pdf")
#
#
# for x in output_latex:
# print(x)
#
#
# twist_learn_method_names = [
# "EBM-One-KL",
# "EBM-One-KL Trained on Only Exact Posterior Samples",
# ]
# make_combined_plot(load_prefixes_toxt, "toxt_bounds_01-14")