decoding results

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+decoding_metrics.png filter=lfs diff=lfs merge=lfs -text

decoding_results.json

@@ -1,53 +1,53 @@
 {
   "summary": {
-    "reference_ppl_mean": 27.4270831861496,
-    "reference_ppl_median": 24.03861141204834,
+    "reference_ppl_mean": 27.426975273132324,
+    "reference_ppl_median": 24.039149284362793,
     "n_samples": 1000,
     "model": "EleutherAI/pythia-160m",
     "max_new_tokens": 70,
     "beam_w1_equals_greedy_frac": 0.998,
-    "reference_judge_ppl_median": 16.601261138916016
+    "reference_judge_ppl_median": 16.60109519958496
   },
   "results": [
-    {
-      "precision": 0.15703556591808848,
-      "recall": 0.10770206338672424,
-      "f1": 0.11695427749203965,
-      "accuracy": 0.013886833130142854,
-      "bleu": 0.4602816499134766,
-      "self_ppl_mean": 2.129318780740721,
-      "self_ppl_median": 1.9777636367502676,
-      "distinct_1": 0.037839837248011834,
-      "distinct_2": 0.11822546571922413,
-      "rep_4gram": 0.3293539897022365,
-      "avg_words": 27.035,
-      "name": "greedy",
-      "strategy": "greedy",
-      "params": {
-        "name": "greedy",
-        "strategy": "greedy"
-      },
-      "time_s": 24.308775424957275,
-      "ms_per_prompt": 24.308775424957275,
-      "gen_tokens_mean": 69.995,
-      "judge_ppl_median": 3.7137224674224854,
-      "judge_ppl_mean": 3.8401719745397567,
-      "bertscore_p": 0.8487334251403809,
-      "bertscore_r": 0.846333384513855,
-      "bertscore_f1": 0.8474068641662598
-    },
     {
       "precision": 0.15680400502615022,
       "recall": 0.10761634910100995,
       "f1": 0.11683061986311886,
       "accuracy": 0.013858261701571426,
       "bleu": 0.45979362180213906,
-      "self_ppl_mean": 2.129485707215857,
-      "self_ppl_median": 1.9777288774584214,
+      "self_ppl_mean": 2.129511192101869,
+      "self_ppl_median": 1.977763465028044,
       "distinct_1": 0.03781326236416057,
       "distinct_2": 0.11821601289629231,
       "rep_4gram": 0.3296551592928797,
       "avg_words": 27.054,
+      "name": "greedy",
+      "strategy": "greedy",
+      "params": {
+        "name": "greedy",
+        "strategy": "greedy"
+      },
+      "time_s": 24.53210139274597,
+      "ms_per_prompt": 24.53210139274597,
+      "gen_tokens_mean": 69.995,
+      "judge_ppl_median": 3.7136964797973633,
+      "judge_ppl_mean": 3.8388092192411425,
+      "bertscore_p": 0.8487208485603333,
+      "bertscore_r": 0.8463250994682312,
+      "bertscore_f1": 0.8473963141441345
+    },
+    {
+      "precision": 0.1568385962211188,
+      "recall": 0.10770206338672424,
+      "f1": 0.11688205526981743,
+      "accuracy": 0.013886833130142854,
+      "bleu": 0.4603869452185167,
+      "self_ppl_mean": 2.1297556273252223,
+      "self_ppl_median": 1.9777561918932842,
+      "distinct_1": 0.03780348102435239,
+      "distinct_2": 0.11810751697939449,
+      "rep_4gram": 0.3296463873630552,
+      "avg_words": 27.061,
       "name": "beam_w1",
       "strategy": "beam",
       "params": {
@@ -55,27 +55,27 @@
         "strategy": "beam",
         "width": 1
       },
-      "time_s": 98.08044266700745,
-      "ms_per_prompt": 98.08044266700745,
+      "time_s": 89.72963857650757,
+      "ms_per_prompt": 89.72963857650757,
       "gen_tokens_mean": 69.995,
-      "judge_ppl_median": 3.7137224674224854,
-      "judge_ppl_mean": 3.838806267619133,
-      "bertscore_p": 0.8487208485603333,
-      "bertscore_r": 0.8463250994682312,
-      "bertscore_f1": 0.8473963141441345
+      "judge_ppl_median": 3.7136964797973633,
+      "judge_ppl_mean": 3.839159709095955,
+      "bertscore_p": 0.8486897945404053,
+      "bertscore_r": 0.8463218808174133,
+      "bertscore_f1": 0.8473787903785706
     },
     {
-      "precision": 0.2013900404889916,
-      "recall": 0.08663449975312773,
-      "f1": 0.11019996943421603,
+      "precision": 0.20124718334613445,
+      "recall": 0.08660508798842184,
+      "f1": 0.11012341441029258,
       "accuracy": 0.010045733414015445,
-      "bleu": 0.2953586002637482,
-      "self_ppl_mean": 1.8974450017258762,
-      "self_ppl_median": 1.8302027765214615,
-      "distinct_1": 0.04578023300333643,
-      "distinct_2": 0.13660822773823988,
-      "rep_4gram": 0.2084343790507725,
-      "avg_words": 18.283,
+      "bleu": 0.29556628142570907,
+      "self_ppl_mean": 1.8974248541153338,
+      "self_ppl_median": 1.8301979892743205,
+      "distinct_1": 0.045782342657342656,
+      "distinct_2": 0.13661581137309292,
+      "rep_4gram": 0.20882754144393487,
+      "avg_words": 18.304,
       "name": "beam_w2",
       "strategy": "beam",
       "params": {
@@ -83,27 +83,27 @@
         "strategy": "beam",
         "width": 2
       },
-      "time_s": 96.86480498313904,
-      "ms_per_prompt": 96.86480498313904,
+      "time_s": 93.67243909835815,
+      "ms_per_prompt": 93.67243909835815,
       "gen_tokens_mean": 69.827,
-      "judge_ppl_median": 4.084466457366943,
-      "judge_ppl_mean": 5.8135339479446415,
-      "bertscore_p": 0.8610000014305115,
-      "bertscore_r": 0.846572756767273,
-      "bertscore_f1": 0.8535789847373962
+      "judge_ppl_median": 4.0844643115997314,
+      "judge_ppl_mean": 5.812758823871612,
+      "bertscore_p": 0.8609820604324341,
+      "bertscore_r": 0.846573531627655,
+      "bertscore_f1": 0.8535703420639038
     },
     {
-      "precision": 0.19775710589874798,
-      "recall": 0.07414467122910813,
-      "f1": 0.09580850458664733,
-      "accuracy": 0.008091172722680333,
-      "bleu": 0.34056160814471953,
-      "self_ppl_mean": 1.7568848200945872,
-      "self_ppl_median": 1.739891141883354,
-      "distinct_1": 0.04018989048929169,
-      "distinct_2": 0.11672463933397959,
-      "rep_4gram": 0.23828927173442535,
-      "avg_words": 18.537,
+      "precision": 0.197717653773812,
+      "recall": 0.07404912467040367,
+      "f1": 0.09570392741330655,
+      "accuracy": 0.008065531697039307,
+      "bleu": 0.3401904436011399,
+      "self_ppl_mean": 1.7576851325086995,
+      "self_ppl_median": 1.739890192921894,
+      "distinct_1": 0.04027860266724259,
+      "distinct_2": 0.11694537982991839,
+      "rep_4gram": 0.23772850694366054,
+      "avg_words": 18.521,
       "name": "beam_w4",
       "strategy": "beam",
       "params": {
@@ -111,27 +111,27 @@
         "strategy": "beam",
         "width": 4
       },
-      "time_s": 105.15537786483765,
-      "ms_per_prompt": 105.15537786483765,
-      "gen_tokens_mean": 69.264,
-      "judge_ppl_median": 3.7111488580703735,
-      "judge_ppl_mean": 549.2648122023344,
-      "bertscore_p": 0.8595147728919983,
-      "bertscore_r": 0.844566285610199,
-      "bertscore_f1": 0.8517965078353882
+      "time_s": 106.44298911094666,
+      "ms_per_prompt": 106.44298911094666,
+      "gen_tokens_mean": 69.253,
+      "judge_ppl_median": 3.722070574760437,
+      "judge_ppl_mean": 549.2117587163449,
+      "bertscore_p": 0.8594975471496582,
+      "bertscore_r": 0.8445566892623901,
+      "bertscore_f1": 0.8517829775810242
     },
     {
-      "precision": 0.20413235594968865,
-      "recall": 0.16095103861563514,
-      "f1": 0.16238850568174357,
-      "accuracy": 0.015531595531122681,
-      "bleu": 0.6557075980512005,
-      "self_ppl_mean": 5.760568807491463,
-      "self_ppl_median": 5.7298509005910425,
-      "distinct_1": 0.07298136645962733,
-      "distinct_2": 0.35367849194920803,
-      "rep_4gram": 0.031188779258196065,
-      "avg_words": 31.556,
+      "precision": 0.20247790936126098,
+      "recall": 0.16240848409656986,
+      "f1": 0.16326782514510493,
+      "accuracy": 0.015387661973703906,
+      "bleu": 0.6321177804113657,
+      "self_ppl_mean": 5.788990184998834,
+      "self_ppl_median": 5.7832007131546685,
+      "distinct_1": 0.07138813761871816,
+      "distinct_2": 0.35044483408013505,
+      "rep_4gram": 0.03276803420190723,
+      "avg_words": 31.798,
       "name": "topk_k5",
       "strategy": "top_k",
       "params": {
@@ -139,27 +139,27 @@
         "strategy": "top_k",
         "k": 5
       },
-      "time_s": 23.9933123588562,
-      "ms_per_prompt": 23.9933123588562,
-      "gen_tokens_mean": 69.674,
-      "judge_ppl_median": 8.357365608215332,
-      "judge_ppl_mean": 10.3706097561121,
-      "bertscore_p": 0.8579887747764587,
-      "bertscore_r": 0.8531241416931152,
-      "bertscore_f1": 0.8554333448410034
+      "time_s": 24.93057608604431,
+      "ms_per_prompt": 24.93057608604431,
+      "gen_tokens_mean": 69.711,
+      "judge_ppl_median": 8.327152252197266,
+      "judge_ppl_mean": 10.00695155286789,
+      "bertscore_p": 0.8572027683258057,
+      "bertscore_r": 0.8528223037719727,
+      "bertscore_f1": 0.8548986911773682
     },
     {
-      "precision": 0.1943437061801331,
-      "recall": 0.17018026211031476,
-      "f1": 0.16424816130281425,
-      "accuracy": 0.013720664726026286,
-      "bleu": 0.4768122146961613,
-      "self_ppl_mean": 10.232085836050464,
-      "self_ppl_median": 9.978154937877655,
-      "distinct_1": 0.08890568670775671,
-      "distinct_2": 0.47079078959187953,
-      "rep_4gram": 0.005152041399905391,
-      "avg_words": 34.396,
+      "precision": 0.18981835217131812,
+      "recall": 0.17505740033246367,
+      "f1": 0.16608450826242258,
+      "accuracy": 0.013520378544886719,
+      "bleu": 0.5262730257755575,
+      "self_ppl_mean": 10.425129562862075,
+      "self_ppl_median": 10.068886870395898,
+      "distinct_1": 0.08757376684110901,
+      "distinct_2": 0.4628797205600252,
+      "rep_4gram": 0.005379857536448407,
+      "avg_words": 35.924,
       "name": "topk_k20",
       "strategy": "top_k",
       "params": {
@@ -167,24 +167,24 @@
         "strategy": "top_k",
         "k": 20
       },
-      "time_s": 23.697141885757446,
-      "ms_per_prompt": 23.697141885757446,
-      "gen_tokens_mean": 69.426,
-      "judge_ppl_median": 14.711297035217285,
-      "judge_ppl_mean": 18.454828331232072
+      "time_s": 24.941646337509155,
+      "ms_per_prompt": 24.941646337509155,
+      "gen_tokens_mean": 69.067,
+      "judge_ppl_median": 14.630431175231934,
+      "judge_ppl_mean": 24.414003211975096
     },
     {
-      "precision": 0.18013398164473904,
-      "recall": 0.16448516354938555,
-      "f1": 0.15464004612237248,
-      "accuracy": 0.011155156675690158,
-      "bleu": 0.22976691112866304,
-      "self_ppl_mean": 14.747323534754505,
-      "self_ppl_median": 13.803862584945907,
-      "distinct_1": 0.10471868305333183,
-      "distinct_2": 0.5331651149975347,
-      "rep_4gram": 0.0014281404078550252,
-      "avg_words": 35.476,
+      "precision": 0.18100045337501575,
+      "recall": 0.16393725762403472,
+      "f1": 0.15556260822373996,
+      "accuracy": 0.010887085233937758,
+      "bleu": 0.40375922428576166,
+      "self_ppl_mean": 14.773549448869355,
+      "self_ppl_median": 13.841562918287725,
+      "distinct_1": 0.10454026270702456,
+      "distinct_2": 0.5344305657604702,
+      "rep_4gram": 0.0027486902217292155,
+      "avg_words": 35.02,
       "name": "topk_k50",
       "strategy": "top_k",
       "params": {
@@ -192,24 +192,24 @@
         "strategy": "top_k",
         "k": 50
       },
-      "time_s": 23.64177703857422,
-      "ms_per_prompt": 23.64177703857422,
-      "gen_tokens_mean": 69.404,
-      "judge_ppl_median": 20.432851791381836,
-      "judge_ppl_mean": 34.98311813139915
+      "time_s": 24.886651039123535,
+      "ms_per_prompt": 24.886651039123535,
+      "gen_tokens_mean": 68.922,
+      "judge_ppl_median": 21.331963539123535,
+      "judge_ppl_mean": 501.72633791828156
     },
     {
-      "precision": 0.20203389091922755,
-      "recall": 0.15151844096638248,
-      "f1": 0.15594146830403016,
-      "accuracy": 0.013403332883272369,
-      "bleu": 0.5574700588317076,
-      "self_ppl_mean": 7.165241852811464,
-      "self_ppl_median": 6.7217895292101915,
-      "distinct_1": 0.09238144153398391,
-      "distinct_2": 0.43317498581962566,
-      "rep_4gram": 0.019412488951243617,
-      "avg_words": 29.205,
+      "precision": 0.20526678420165223,
+      "recall": 0.15309069901051064,
+      "f1": 0.15840480805376173,
+      "accuracy": 0.014216155649643127,
+      "bleu": 0.50328331736766,
+      "self_ppl_mean": 7.068113830843848,
+      "self_ppl_median": 6.912323629738417,
+      "distinct_1": 0.09306409130816505,
+      "distinct_2": 0.43222086318364494,
+      "rep_4gram": 0.02030865920372129,
+      "avg_words": 29.614,
       "name": "topk_k50_t0.7",
       "strategy": "top_k",
       "params": {
@@ -218,24 +218,24 @@
         "k": 50,
         "temperature": 0.7
       },
-      "time_s": 23.68122911453247,
-      "ms_per_prompt": 23.68122911453247,
-      "gen_tokens_mean": 69.607,
-      "judge_ppl_median": 10.626161575317383,
-      "judge_ppl_mean": 17.331569744825362
+      "time_s": 24.90457844734192,
+      "ms_per_prompt": 24.90457844734192,
+      "gen_tokens_mean": 69.718,
+      "judge_ppl_median": 10.698809146881104,
+      "judge_ppl_mean": 15.45937124478817
     },
     {
-      "precision": 0.20909230371572832,
-      "recall": 0.14641056202688646,
-      "f1": 0.15683576868798077,
-      "accuracy": 0.015255414154035243,
-      "bleu": 0.29849302912542913,
-      "self_ppl_mean": 6.067967456491997,
-      "self_ppl_median": 5.979495040329027,
-      "distinct_1": 0.09776068923777992,
-      "distinct_2": 0.4221270626453542,
-      "rep_4gram": 0.027907301032833808,
-      "avg_words": 28.089,
+      "precision": 0.20739706440864433,
+      "recall": 0.1499013118788447,
+      "f1": 0.15918566183599664,
+      "accuracy": 0.014822320849886555,
+      "bleu": 0.4478536424658865,
+      "self_ppl_mean": 6.073593265666012,
+      "self_ppl_median": 5.982498789525279,
+      "distinct_1": 0.09500420993544766,
+      "distinct_2": 0.41255817335660266,
+      "rep_4gram": 0.03136431125812165,
+      "avg_words": 28.504,
       "name": "topp_p0.5",
       "strategy": "top_p",
       "params": {
@@ -243,24 +243,24 @@
         "strategy": "top_p",
         "p": 0.5
       },
-      "time_s": 25.245723962783813,
-      "ms_per_prompt": 25.245723962783813,
-      "gen_tokens_mean": 69.876,
-      "judge_ppl_median": 9.52494764328003,
-      "judge_ppl_mean": 12.933369012594223
+      "time_s": 27.650500059127808,
+      "ms_per_prompt": 27.650500059127808,
+      "gen_tokens_mean": 69.866,
+      "judge_ppl_median": 9.648499011993408,
+      "judge_ppl_mean": 13.219340204000472
     },
     {
-      "precision": 0.1790151623639173,
-      "recall": 0.15383583830048134,
-      "f1": 0.14916733197352083,
-      "accuracy": 0.01169844779025957,
-      "bleu": 0.21170796316754,
-      "self_ppl_mean": 16.263590068968764,
-      "self_ppl_median": 14.61157853686635,
-      "distinct_1": 0.1348369399573301,
-      "distinct_2": 0.5826313804375157,
-      "rep_4gram": 0.002888150162719001,
-      "avg_words": 32.81,
+      "precision": 0.1786455293948896,
+      "recall": 0.15898556161207772,
+      "f1": 0.15182183169853083,
+      "accuracy": 0.01125190065874159,
+      "bleu": 0.22743045693599412,
+      "self_ppl_mean": 16.296575027955598,
+      "self_ppl_median": 14.50375505361637,
+      "distinct_1": 0.13724910579680155,
+      "distinct_2": 0.5839322694603484,
+      "rep_4gram": 0.003038437207161451,
+      "avg_words": 33.829,
       "name": "topp_p0.8",
       "strategy": "top_p",
       "params": {
@@ -268,24 +268,24 @@
         "strategy": "top_p",
         "p": 0.8
       },
-      "time_s": 25.11012601852417,
-      "ms_per_prompt": 25.11012601852417,
-      "gen_tokens_mean": 69.23,
-      "judge_ppl_median": 23.586853981018066,
-      "judge_ppl_mean": 8093.695829333544
+      "time_s": 27.59784722328186,
+      "ms_per_prompt": 27.59784722328186,
+      "gen_tokens_mean": 69.023,
+      "judge_ppl_median": 23.389395713806152,
+      "judge_ppl_mean": 4002.7205463142395
     },
     {
-      "precision": 0.15996633836782667,
-      "recall": 0.1436485453422705,
-      "f1": 0.1358491321482202,
-      "accuracy": 0.00975848890740479,
-      "bleu": 0.15980548317421198,
-      "self_ppl_mean": 38.624090383453975,
-      "self_ppl_median": 32.174860783175475,
-      "distinct_1": 0.1859961430054888,
-      "distinct_2": 0.6953437891711761,
-      "rep_4gram": 0.0004418966228079527,
-      "avg_words": 33.705,
+      "precision": 0.15578726318266997,
+      "recall": 0.14156256433661007,
+      "f1": 0.13438127406575798,
+      "accuracy": 0.008718096877912114,
+      "bleu": 0.16851247100645653,
+      "self_ppl_mean": 40.52525353706895,
+      "self_ppl_median": 34.312852488083806,
+      "distinct_1": 0.18865485737786786,
+      "distinct_2": 0.7045688350983359,
+      "rep_4gram": 0.0005719175842843666,
+      "avg_words": 34.041,
       "name": "topp_p0.95",
       "strategy": "top_p",
       "params": {
@@ -293,24 +293,24 @@
         "strategy": "top_p",
         "p": 0.95
       },
-      "time_s": 25.150532722473145,
-      "ms_per_prompt": 25.150532722473145,
-      "gen_tokens_mean": 68.846,
-      "judge_ppl_median": 54.63137245178223,
-      "judge_ppl_mean": 3939.534708950758
+      "time_s": 27.591065883636475,
+      "ms_per_prompt": 27.591065883636475,
+      "gen_tokens_mean": 68.753,
+      "judge_ppl_median": 58.844425201416016,
+      "judge_ppl_mean": 4085.3288768868447
     },
     {
-      "precision": 0.2018115838286414,
-      "recall": 0.1429945363511889,
-      "f1": 0.15115617299132603,
-      "accuracy": 0.013776023575061398,
-      "bleu": 0.557218536514725,
-      "self_ppl_mean": 5.709627802185269,
-      "self_ppl_median": 5.569257784916102,
-      "distinct_1": 0.09286149654010255,
-      "distinct_2": 0.4019261517867103,
-      "rep_4gram": 0.038554351907382906,
-      "avg_words": 27.891,
+      "precision": 0.20651589086843022,
+      "recall": 0.1484358064877369,
+      "f1": 0.15685438845860386,
+      "accuracy": 0.014285365431144705,
+      "bleu": 0.6053996783877227,
+      "self_ppl_mean": 5.729693244652217,
+      "self_ppl_median": 5.577883722137798,
+      "distinct_1": 0.09334971047552203,
+      "distinct_2": 0.4059283506092017,
+      "rep_4gram": 0.03475423802329605,
+      "avg_words": 28.495,
       "name": "topp_p0.9_t0.7",
       "strategy": "top_p",
       "params": {
@@ -319,92 +319,92 @@
         "p": 0.9,
         "temperature": 0.7
       },
-      "time_s": 25.12675452232361,
-      "ms_per_prompt": 25.12675452232361,
-      "gen_tokens_mean": 69.744,
-      "judge_ppl_median": 9.09609079360962,
-      "judge_ppl_mean": 12.747726317882538
+      "time_s": 27.645728826522827,
+      "ms_per_prompt": 27.645728826522827,
+      "gen_tokens_mean": 69.939,
+      "judge_ppl_median": 9.250245571136475,
+      "judge_ppl_mean": 11.254222588062287
     }
   ],
   "beam_timing": [
     {
       "mode": "natural",
       "width": 1,
-      "mean_s_per_prompt": 0.7000857973098755,
-      "std_s": 0.03165861046389621,
-      "mean_ms_per_token": 10.001225675855366,
+      "mean_s_per_prompt": 0.6848450946807861,
+      "std_s": 0.03387934526472246,
+      "mean_ms_per_token": 9.78350135258266,
       "mean_output_tokens": 70.0,
-      "total_s": 70.00857973098755,
+      "total_s": 68.48450946807861,
       "n_prompts": 100
     },
     {
       "mode": "natural",
       "width": 2,
-      "mean_s_per_prompt": 0.7082566213607788,
-      "std_s": 0.010652223812277732,
-      "mean_ms_per_token": 10.138293554772547,
+      "mean_s_per_prompt": 0.6798255443572998,
+      "std_s": 0.009905260321141293,
+      "mean_ms_per_token": 9.7313968216702,
       "mean_output_tokens": 69.86,
-      "total_s": 70.82566213607788,
+      "total_s": 67.98255443572998,
       "n_prompts": 100
     },
     {
       "mode": "natural",
       "width": 4,
-      "mean_s_per_prompt": 0.7128961229324341,
-      "std_s": 0.02926986205734486,
-      "mean_ms_per_token": 10.244703469472553,
+      "mean_s_per_prompt": 0.6822133040428162,
+      "std_s": 0.027737611702376378,
+      "mean_ms_per_token": 9.803958435890708,
       "mean_output_tokens": 69.59,
-      "total_s": 71.28961229324341,
+      "total_s": 68.22133040428162,
       "n_prompts": 100
     },
     {
       "mode": "natural",
       "width": 8,
-      "mean_s_per_prompt": 0.7481088328361511,
-      "std_s": 0.03363774855082431,
-      "mean_ms_per_token": 10.809140646212507,
+      "mean_s_per_prompt": 0.7037396359443665,
+      "std_s": 0.03137733007915173,
+      "mean_ms_per_token": 10.1683868747084,
       "mean_output_tokens": 69.34,
-      "total_s": 74.81088328361511,
+      "total_s": 70.37396359443665,
       "n_prompts": 100
     },
     {
       "mode": "fixed",
       "width": 1,
-      "mean_s_per_prompt": 0.693102993965149,
-      "std_s": 0.0013031337825710797,
-      "mean_ms_per_token": 9.90147134235927,
+      "mean_s_per_prompt": 0.6784237551689148,
+      "std_s": 0.0016587773214447714,
+      "mean_ms_per_token": 9.691767930984499,
       "mean_output_tokens": 70.0,
-      "total_s": 69.31029939651489,
+      "total_s": 67.84237551689148,
       "n_prompts": 100
     },
     {
       "mode": "fixed",
       "width": 2,
-      "mean_s_per_prompt": 0.7096165156364441,
-      "std_s": 0.0011025559183209961,
-      "mean_ms_per_token": 10.137378794806345,
+      "mean_s_per_prompt": 0.6811458969116211,
+      "std_s": 0.001066027939126381,
+      "mean_ms_per_token": 9.730655670166016,
       "mean_output_tokens": 70.0,
-      "total_s": 70.96165156364441,
+      "total_s": 68.11458969116211,
       "n_prompts": 100
     },
     {
       "mode": "fixed",
       "width": 4,
-      "mean_s_per_prompt": 0.7169742965698243,
-      "std_s": 0.0014611031573721494,
-      "mean_ms_per_token": 10.24248995099749,
+      "mean_s_per_prompt": 0.6865559673309326,
+      "std_s": 0.0012468601240148754,
+      "mean_ms_per_token": 9.807942390441896,
       "mean_output_tokens": 70.0,
-      "total_s": 71.69742965698242,
+      "total_s": 68.65559673309326,
       "n_prompts": 100
     },
     {
       "mode": "fixed",
       "width": 8,
-      "mean_s_per_prompt": 0.754322988986969,
-      "std_s": 0.0015383658998336079,
-      "mean_ms_per_token": 10.776042699813841,
+      "mean_s_per_prompt": 0.7092588877677918,
+      "std_s": 0.0009664558877848735,
+      "mean_ms_per_token": 10.132269825254168,
       "mean_output_tokens": 70.0,
-      "total_s": 75.4322988986969,
+      "total_s": 70.92588877677917,
       "n_prompts": 100
     }
   ],
@@ -495,6 +495,9 @@
     "hf_user": null,
     "repo_name": "anlp-a2-decoding",
     "push_to_hub": true,
-    "private": false
+    "private": false,
+    "wandb": true,
+    "wandb_project": "anlp-a2",
+    "wandb_entity": null
   }
 }

decoding_table.csv

@@ -0,0 +1,13 @@
+name,accuracy,precision,recall,f1,bleu,self_ppl_median,judge_ppl_median,bertscore_f1,distinct_2,rep_4gram,avg_words,ms_per_prompt
+greedy,0.013886833130142854,0.15703556591808848,0.10770206338672424,0.11695427749203965,0.4602816499134766,1.9777636367502676,3.7137224674224854,0.8474068641662598,0.11822546571922413,0.3293539897022365,27.035,24.308775424957275
+beam_w1,0.013858261701571426,0.15680400502615022,0.10761634910100995,0.11683061986311886,0.45979362180213906,1.9777288774584214,3.7137224674224854,0.8473963141441345,0.11821601289629231,0.3296551592928797,27.054,98.08044266700745
+beam_w2,0.010045733414015445,0.2013900404889916,0.08663449975312773,0.11019996943421603,0.2953586002637482,1.8302027765214615,4.084466457366943,0.8535789847373962,0.13660822773823988,0.2084343790507725,18.283,96.86480498313904
+beam_w4,0.008091172722680333,0.19775710589874798,0.07414467122910813,0.09580850458664733,0.34056160814471953,1.739891141883354,3.7111488580703735,0.8517965078353882,0.11672463933397959,0.23828927173442535,18.537,105.15537786483765
+topk_k5,0.015531595531122681,0.20413235594968865,0.16095103861563514,0.16238850568174357,0.6557075980512005,5.7298509005910425,8.357365608215332,0.8554333448410034,0.35367849194920803,0.031188779258196065,31.556,23.9933123588562
+topk_k20,0.013720664726026286,0.1943437061801331,0.17018026211031476,0.16424816130281425,0.4768122146961613,9.978154937877655,14.711297035217285,,0.47079078959187953,0.005152041399905391,34.396,23.697141885757446
+topk_k50,0.011155156675690158,0.18013398164473904,0.16448516354938555,0.15464004612237248,0.22976691112866304,13.803862584945907,20.432851791381836,,0.5331651149975347,0.0014281404078550252,35.476,23.64177703857422
+topk_k50_t0.7,0.013403332883272369,0.20203389091922755,0.15151844096638248,0.15594146830403016,0.5574700588317076,6.7217895292101915,10.626161575317383,,0.43317498581962566,0.019412488951243617,29.205,23.68122911453247
+topp_p0.5,0.015255414154035243,0.20909230371572832,0.14641056202688646,0.15683576868798077,0.29849302912542913,5.979495040329027,9.52494764328003,,0.4221270626453542,0.027907301032833808,28.089,25.245723962783813
+topp_p0.8,0.01169844779025957,0.1790151623639173,0.15383583830048134,0.14916733197352083,0.21170796316754,14.61157853686635,23.586853981018066,,0.5826313804375157,0.002888150162719001,32.81,25.11012601852417
+topp_p0.95,0.00975848890740479,0.15996633836782667,0.1436485453422705,0.1358491321482202,0.15980548317421198,32.174860783175475,54.63137245178223,,0.6953437891711761,0.0004418966228079527,33.705,25.150532722473145
+topp_p0.9_t0.7,0.013776023575061398,0.2018115838286414,0.1429945363511889,0.15115617299132603,0.557218536514725,5.569257784916102,9.09609079360962,,0.4019261517867103,0.038554351907382906,27.891,25.12675452232361