Model save

README.md

@@ -2,15 +2,11 @@
 license: apache-2.0
 base_model: mistralai/Mistral-7B-Instruct-v0.2
 tags:
-- alignment-handbook
-- trl
-- sft
-- generated_from_trainer
 - trl
 - sft
 - generated_from_trainer
 datasets:
-- preference-data
+- generator
 model-index:
 - name: preference_0.1p_seed42_level3_rare
   results: []
@@ -21,9 +17,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # preference_0.1p_seed42_level3_rare
 
-This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the preference-data dataset.
+This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the generator dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.7365
+- Loss: 0.7390
 
 ## Model description
 
@@ -60,7 +56,7 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 0.5575        | 0.9967 | 229  | 0.7365          |
+| 0.6207        | 0.9979 | 234  | 0.7390          |
 
 
 ### Framework versions

all_results.json

@@ -1,14 +1,9 @@
 {
-    "epoch": 0.9967355821545157,
-    "eval_loss": 0.7364564538002014,
-    "eval_runtime": 106.4286,
-    "eval_samples": 999,
-    "eval_samples_per_second": 3.664,
-    "eval_steps_per_second": 0.921,
-    "total_flos": 47895596236800.0,
-    "train_loss": 0.7106172507506792,
-    "train_runtime": 7254.3633,
+    "epoch": 0.997867803837953,
+    "total_flos": 48942494515200.0,
+    "train_loss": 0.7695284368645432,
+    "train_runtime": 7306.9109,
     "train_samples": 18788,
-    "train_samples_per_second": 1.013,
+    "train_samples_per_second": 1.026,
     "train_steps_per_second": 0.032
 }

config.json

@@ -22,6 +22,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.43.4",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 32000
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d48bf1eb9b91e3e33ef4b68df0b104dc5ed8a1c29489fdf24d1112e53c1873dc
+oid sha256:86f4d7f7f71205d776a15d9bb271ef8e4c5b85a4464a7d9cc8fabf5abbe6742a
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2e966f0e6454b578cc2ce0bfc4e1d4d089ffa63860cbff54dd59cc1a4bea67a6
+oid sha256:8d3da52201a469e680648b0686698f7ed8cc2c1f0cf1fb25f271d2e05cdf9b69
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f001e5166a1403d88df7fb44a8b577563bfa911db02dc5cb3a05a62420984129
+oid sha256:e053a2e5d6ca1464762eb049d3dfabc9f0e6457f3bc4384424e2868e4071f4ba
 size 4540516344

runs/Aug13_13-34-23_COE-CS-sv004/events.out.tfevents.1723556233.COE-CS-sv004.2672049.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af08c6c58d2a3c6ef9cf9bc2649171c3215d1f8b4368f5fd3012ccbca822bf94
+size 15669

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 0.9967355821545157,
-    "total_flos": 47895596236800.0,
-    "train_loss": 0.7106172507506792,
-    "train_runtime": 7254.3633,
+    "epoch": 0.997867803837953,
+    "total_flos": 48942494515200.0,
+    "train_loss": 0.7695284368645432,
+    "train_runtime": 7306.9109,
     "train_samples": 18788,
-    "train_samples_per_second": 1.013,
+    "train_samples_per_second": 1.026,
     "train_steps_per_second": 0.032
 }

trainer_state.json

@@ -1,355 +1,362 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9967355821545157,
+  "epoch": 0.997867803837953,
   "eval_steps": 500,
-  "global_step": 229,
+  "global_step": 234,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.004352557127312296,
-      "grad_norm": 23.969777401039607,
-      "learning_rate": 4.347826086956522e-07,
-      "loss": 1.4137,
+      "epoch": 0.0042643923240938165,
+      "grad_norm": 25.654364462455888,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 1.5118,
       "step": 1
     },
     {
-      "epoch": 0.02176278563656148,
-      "grad_norm": 8.67616469459428,
-      "learning_rate": 2.173913043478261e-06,
-      "loss": 1.3829,
+      "epoch": 0.021321961620469083,
+      "grad_norm": 9.217679826848354,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 1.4588,
       "step": 5
     },
     {
-      "epoch": 0.04352557127312296,
-      "grad_norm": 3.5647414259572265,
-      "learning_rate": 4.347826086956522e-06,
-      "loss": 1.1134,
+      "epoch": 0.042643923240938165,
+      "grad_norm": 3.8364568135386343,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.1844,
       "step": 10
     },
     {
-      "epoch": 0.06528835690968444,
-      "grad_norm": 3.2984928713958666,
-      "learning_rate": 6.521739130434783e-06,
-      "loss": 0.9599,
+      "epoch": 0.06396588486140725,
+      "grad_norm": 2.7685426084723606,
+      "learning_rate": 6.25e-06,
+      "loss": 1.0266,
       "step": 15
     },
     {
-      "epoch": 0.08705114254624592,
-      "grad_norm": 2.725944554945545,
-      "learning_rate": 8.695652173913044e-06,
-      "loss": 0.924,
+      "epoch": 0.08528784648187633,
+      "grad_norm": 2.714167254334698,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.9764,
       "step": 20
     },
     {
-      "epoch": 0.1088139281828074,
-      "grad_norm": 2.690177981501858,
-      "learning_rate": 9.997674418116759e-06,
-      "loss": 0.8916,
+      "epoch": 0.10660980810234541,
+      "grad_norm": 2.575043424767955,
+      "learning_rate": 9.999440509051367e-06,
+      "loss": 0.9474,
       "step": 25
     },
     {
-      "epoch": 0.1305767138193689,
-      "grad_norm": 2.575585422886906,
-      "learning_rate": 9.971536460096021e-06,
-      "loss": 0.8802,
+      "epoch": 0.1279317697228145,
+      "grad_norm": 2.5473126644658635,
+      "learning_rate": 9.979871469976197e-06,
+      "loss": 0.9265,
       "step": 30
     },
     {
-      "epoch": 0.15233949945593037,
-      "grad_norm": 2.4168957628954746,
-      "learning_rate": 9.916505976821262e-06,
-      "loss": 0.8682,
+      "epoch": 0.14925373134328357,
+      "grad_norm": 2.7728522540985927,
+      "learning_rate": 9.932452969617607e-06,
+      "loss": 0.9103,
       "step": 35
     },
     {
-      "epoch": 0.17410228509249184,
-      "grad_norm": 2.597564344293517,
-      "learning_rate": 9.832902782828801e-06,
-      "loss": 0.8501,
+      "epoch": 0.17057569296375266,
+      "grad_norm": 2.262656802643975,
+      "learning_rate": 9.857450191464337e-06,
+      "loss": 0.9089,
       "step": 40
     },
     {
-      "epoch": 0.19586507072905332,
-      "grad_norm": 2.8200473310088534,
-      "learning_rate": 9.721212745498493e-06,
-      "loss": 0.8429,
+      "epoch": 0.19189765458422176,
+      "grad_norm": 2.2929860595064353,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 0.8839,
       "step": 45
     },
     {
-      "epoch": 0.2176278563656148,
-      "grad_norm": 2.528361018534502,
-      "learning_rate": 9.582084961392358e-06,
-      "loss": 0.8167,
+      "epoch": 0.21321961620469082,
+      "grad_norm": 2.9962187125117556,
+      "learning_rate": 9.626521502369984e-06,
+      "loss": 0.8779,
       "step": 50
     },
     {
-      "epoch": 0.23939064200217627,
-      "grad_norm": 2.3876192230621047,
-      "learning_rate": 9.416327983972304e-06,
-      "loss": 0.8087,
+      "epoch": 0.2345415778251599,
+      "grad_norm": 2.4461853196937744,
+      "learning_rate": 9.471887038331686e-06,
+      "loss": 0.8655,
       "step": 55
     },
     {
-      "epoch": 0.2611534276387378,
-      "grad_norm": 2.4893211482371767,
-      "learning_rate": 9.22490512461995e-06,
-      "loss": 0.8027,
+      "epoch": 0.255863539445629,
+      "grad_norm": 2.548713200713329,
+      "learning_rate": 9.292243968009332e-06,
+      "loss": 0.8452,
       "step": 60
     },
     {
-      "epoch": 0.28291621327529926,
-      "grad_norm": 2.958049999378203,
-      "learning_rate": 9.008928854267054e-06,
-      "loss": 0.7661,
+      "epoch": 0.2771855010660981,
+      "grad_norm": 2.354080355646257,
+      "learning_rate": 9.088596928322158e-06,
+      "loss": 0.8453,
       "step": 65
     },
     {
-      "epoch": 0.30467899891186073,
-      "grad_norm": 3.0632090846251523,
-      "learning_rate": 8.769654338171986e-06,
-      "loss": 0.7773,
+      "epoch": 0.29850746268656714,
+      "grad_norm": 2.3350186621937494,
+      "learning_rate": 8.862084796122998e-06,
+      "loss": 0.8213,
       "step": 70
     },
     {
-      "epoch": 0.3264417845484222,
-      "grad_norm": 2.6123024349921926,
-      "learning_rate": 8.508472141415468e-06,
-      "loss": 0.7753,
+      "epoch": 0.31982942430703626,
+      "grad_norm": 2.352888208422696,
+      "learning_rate": 8.613974319136959e-06,
+      "loss": 0.8087,
       "step": 75
     },
     {
-      "epoch": 0.3482045701849837,
-      "grad_norm": 2.695553953076791,
-      "learning_rate": 8.226900147508205e-06,
-      "loss": 0.7504,
+      "epoch": 0.3411513859275053,
+      "grad_norm": 2.626490865987853,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 0.8,
       "step": 80
     },
     {
-      "epoch": 0.36996735582154516,
-      "grad_norm": 3.1757782840030604,
-      "learning_rate": 7.92657473707621e-06,
-      "loss": 0.7358,
+      "epoch": 0.3624733475479744,
+      "grad_norm": 2.2564126156464934,
+      "learning_rate": 8.058621495575032e-06,
+      "loss": 0.7883,
       "step": 85
     },
     {
-      "epoch": 0.39173014145810664,
-      "grad_norm": 2.4347947902045326,
-      "learning_rate": 7.609241277889583e-06,
-      "loss": 0.738,
+      "epoch": 0.3837953091684435,
+      "grad_norm": 2.536678489630529,
+      "learning_rate": 7.754484907260513e-06,
+      "loss": 0.7797,
       "step": 90
     },
     {
-      "epoch": 0.4134929270946681,
-      "grad_norm": 2.4759060035072924,
-      "learning_rate": 7.276743981502856e-06,
-      "loss": 0.7013,
+      "epoch": 0.4051172707889126,
+      "grad_norm": 2.330261835490306,
+      "learning_rate": 7.434944122021837e-06,
+      "loss": 0.7704,
       "step": 95
     },
     {
-      "epoch": 0.4352557127312296,
-      "grad_norm": 2.4837499004026506,
-      "learning_rate": 6.931015185455915e-06,
-      "loss": 0.7042,
+      "epoch": 0.42643923240938164,
+      "grad_norm": 2.375473887900136,
+      "learning_rate": 7.101786141547829e-06,
+      "loss": 0.7491,
       "step": 100
     },
     {
-      "epoch": 0.45701849836779107,
-      "grad_norm": 2.589964756825322,
-      "learning_rate": 6.574064123322925e-06,
-      "loss": 0.6864,
+      "epoch": 0.44776119402985076,
+      "grad_norm": 2.31845485895562,
+      "learning_rate": 6.7568741204067145e-06,
+      "loss": 0.7426,
       "step": 105
     },
     {
-      "epoch": 0.47878128400435255,
-      "grad_norm": 2.4839924992380187,
-      "learning_rate": 6.207965247873151e-06,
-      "loss": 0.6839,
+      "epoch": 0.4690831556503198,
+      "grad_norm": 2.2326175780721513,
+      "learning_rate": 6.402136946530014e-06,
+      "loss": 0.7366,
       "step": 110
     },
     {
-      "epoch": 0.500544069640914,
-      "grad_norm": 2.470265064900677,
-      "learning_rate": 5.834846175204612e-06,
-      "loss": 0.6657,
+      "epoch": 0.4904051172707889,
+      "grad_norm": 2.444799836226394,
+      "learning_rate": 6.039558454088796e-06,
+      "loss": 0.7294,
       "step": 115
     },
     {
-      "epoch": 0.5223068552774756,
-      "grad_norm": 2.374809615056107,
-      "learning_rate": 5.456875319914355e-06,
-      "loss": 0.6616,
+      "epoch": 0.511727078891258,
+      "grad_norm": 2.42023799653421,
+      "learning_rate": 5.671166329088278e-06,
+      "loss": 0.7346,
       "step": 120
     },
     {
-      "epoch": 0.544069640914037,
-      "grad_norm": 2.546663639426302,
-      "learning_rate": 5.07624929316463e-06,
-      "loss": 0.641,
+      "epoch": 0.5330490405117271,
+      "grad_norm": 2.525769921790198,
+      "learning_rate": 5.299020769725172e-06,
+      "loss": 0.716,
       "step": 125
     },
     {
-      "epoch": 0.5658324265505985,
-      "grad_norm": 2.4876523628247615,
-      "learning_rate": 4.6951801368822055e-06,
-      "loss": 0.6407,
+      "epoch": 0.5543710021321961,
+      "grad_norm": 2.210624855154462,
+      "learning_rate": 4.9252029649236835e-06,
+      "loss": 0.7087,
       "step": 130
     },
     {
-      "epoch": 0.5875952121871599,
-      "grad_norm": 2.4763298540381666,
-      "learning_rate": 4.31588246828045e-06,
-      "loss": 0.6324,
+      "epoch": 0.5756929637526652,
+      "grad_norm": 2.260417777455262,
+      "learning_rate": 4.551803455482833e-06,
+      "loss": 0.6979,
       "step": 135
     },
     {
-      "epoch": 0.6093579978237215,
-      "grad_norm": 2.776813210102937,
-      "learning_rate": 3.940560609414894e-06,
-      "loss": 0.6377,
+      "epoch": 0.5970149253731343,
+      "grad_norm": 2.5410734519213847,
+      "learning_rate": 4.180910442924312e-06,
+      "loss": 0.6758,
       "step": 140
     },
     {
-      "epoch": 0.6311207834602829,
-      "grad_norm": 2.35611378965622,
-      "learning_rate": 3.5713957765700224e-06,
-      "loss": 0.6239,
+      "epoch": 0.6183368869936035,
+      "grad_norm": 2.2197214614990983,
+      "learning_rate": 3.8145981114225135e-06,
+      "loss": 0.6832,
       "step": 145
     },
     {
-      "epoch": 0.6528835690968444,
-      "grad_norm": 2.2743328506833427,
-      "learning_rate": 3.2105334039272924e-06,
-      "loss": 0.5998,
+      "epoch": 0.6396588486140725,
+      "grad_norm": 2.417478197312417,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.6705,
       "step": 150
     },
     {
-      "epoch": 0.6746463547334058,
-      "grad_norm": 2.9956954866670285,
-      "learning_rate": 2.860070675184036e-06,
-      "loss": 0.5962,
+      "epoch": 0.6609808102345416,
+      "grad_norm": 2.193206874567919,
+      "learning_rate": 3.1038726867353587e-06,
+      "loss": 0.6909,
       "step": 155
     },
     {
-      "epoch": 0.6964091403699674,
-      "grad_norm": 2.595628719529078,
-      "learning_rate": 2.522044335584322e-06,
-      "loss": 0.6027,
+      "epoch": 0.6823027718550106,
+      "grad_norm": 2.3141978562259133,
+      "learning_rate": 2.7634342584218364e-06,
+      "loss": 0.678,
       "step": 160
     },
     {
-      "epoch": 0.7181719260065288,
-      "grad_norm": 2.671379662704793,
-      "learning_rate": 2.1984188551932513e-06,
-      "loss": 0.593,
+      "epoch": 0.7036247334754797,
+      "grad_norm": 2.20282691421215,
+      "learning_rate": 2.43550361297047e-06,
+      "loss": 0.6646,
       "step": 165
     },
     {
-      "epoch": 0.7399347116430903,
-      "grad_norm": 2.31173255024942,
-      "learning_rate": 1.8910750122048638e-06,
-      "loss": 0.5826,
+      "epoch": 0.7249466950959488,
+      "grad_norm": 2.3241432733966962,
+      "learning_rate": 2.1219146715716332e-06,
+      "loss": 0.6633,
       "step": 170
     },
     {
-      "epoch": 0.7616974972796517,
-      "grad_norm": 2.320313991995028,
-      "learning_rate": 1.601798962632799e-06,
-      "loss": 0.564,
+      "epoch": 0.746268656716418,
+      "grad_norm": 2.3658483418520464,
+      "learning_rate": 1.8244211507891064e-06,
+      "loss": 0.6516,
       "step": 175
     },
     {
-      "epoch": 0.7834602829162133,
-      "grad_norm": 2.3880739142797864,
-      "learning_rate": 1.3322718599061252e-06,
-      "loss": 0.5675,
+      "epoch": 0.767590618336887,
+      "grad_norm": 2.259696417637488,
+      "learning_rate": 1.544686755065677e-06,
+      "loss": 0.6418,
       "step": 180
     },
     {
-      "epoch": 0.8052230685527747,
-      "grad_norm": 2.350078693554661,
-      "learning_rate": 1.0840600846970333e-06,
-      "loss": 0.5643,
+      "epoch": 0.7889125799573561,
+      "grad_norm": 2.284368925546414,
+      "learning_rate": 1.2842758726130283e-06,
+      "loss": 0.6405,
       "step": 185
     },
     {
-      "epoch": 0.8269858541893362,
-      "grad_norm": 2.2922329677792286,
-      "learning_rate": 8.586061417605668e-07,
-      "loss": 0.5686,
+      "epoch": 0.8102345415778252,
+      "grad_norm": 2.2174015564488223,
+      "learning_rate": 1.044644826718295e-06,
+      "loss": 0.6359,
       "step": 190
     },
     {
-      "epoch": 0.8487486398258978,
-      "grad_norm": 2.2238743664815086,
-      "learning_rate": 6.572202766902569e-07,
-      "loss": 0.5542,
+      "epoch": 0.8315565031982942,
+      "grad_norm": 2.3098966859462076,
+      "learning_rate": 8.271337313934869e-07,
+      "loss": 0.6232,
       "step": 195
     },
     {
-      "epoch": 0.8705114254624592,
-      "grad_norm": 2.453849740180281,
-      "learning_rate": 4.81072861309591e-07,
-      "loss": 0.5717,
+      "epoch": 0.8528784648187633,
+      "grad_norm": 2.240425165408693,
+      "learning_rate": 6.329589969143518e-07,
+      "loss": 0.6263,
       "step": 200
     },
     {
-      "epoch": 0.8922742110990207,
-      "grad_norm": 2.3379496520696197,
-      "learning_rate": 3.3118759195232273e-07,
-      "loss": 0.5505,
+      "epoch": 0.8742004264392325,
+      "grad_norm": 2.203409177091297,
+      "learning_rate": 4.632065271606756e-07,
+      "loss": 0.6299,
       "step": 205
     },
     {
-      "epoch": 0.9140369967355821,
-      "grad_norm": 2.3452550178883507,
-      "learning_rate": 2.0843554016039326e-07,
-      "loss": 0.5496,
+      "epoch": 0.8955223880597015,
+      "grad_norm": 2.1702011902470724,
+      "learning_rate": 3.18825646801314e-07,
+      "loss": 0.636,
       "step": 210
     },
     {
-      "epoch": 0.9357997823721437,
-      "grad_norm": 2.279878770276356,
-      "learning_rate": 1.1353009037437523e-07,
-      "loss": 0.5506,
+      "epoch": 0.9168443496801706,
+      "grad_norm": 2.202446820245564,
+      "learning_rate": 2.006237922855553e-07,
+      "loss": 0.6182,
       "step": 215
     },
     {
-      "epoch": 0.9575625680087051,
-      "grad_norm": 2.2351766119106924,
-      "learning_rate": 4.702279403650534e-08,
-      "loss": 0.5547,
+      "epoch": 0.9381663113006397,
+      "grad_norm": 2.118840248626809,
+      "learning_rate": 1.0926199633097156e-07,
+      "loss": 0.609,
       "step": 220
     },
     {
-      "epoch": 0.9793253536452666,
-      "grad_norm": 2.2925674952536825,
-      "learning_rate": 9.300164200530815e-09,
-      "loss": 0.5575,
+      "epoch": 0.9594882729211087,
+      "grad_norm": 2.168175873632397,
+      "learning_rate": 4.52511911603265e-08,
+      "loss": 0.6173,
       "step": 225
     },
     {
-      "epoch": 0.9967355821545157,
-      "eval_loss": 0.7364564538002014,
-      "eval_runtime": 106.3104,
-      "eval_samples_per_second": 3.669,
+      "epoch": 0.9808102345415778,
+      "grad_norm": 2.2624619803066617,
+      "learning_rate": 8.949351161324227e-09,
+      "loss": 0.6207,
+      "step": 230
+    },
+    {
+      "epoch": 0.997867803837953,
+      "eval_loss": 0.7390011548995972,
+      "eval_runtime": 106.273,
+      "eval_samples_per_second": 3.67,
       "eval_steps_per_second": 0.922,
-      "step": 229
+      "step": 234
     },
     {
-      "epoch": 0.9967355821545157,
-      "step": 229,
-      "total_flos": 47895596236800.0,
-      "train_loss": 0.7106172507506792,
-      "train_runtime": 7254.3633,
-      "train_samples_per_second": 1.013,
+      "epoch": 0.997867803837953,
+      "step": 234,
+      "total_flos": 48942494515200.0,
+      "train_loss": 0.7695284368645432,
+      "train_runtime": 7306.9109,
+      "train_samples_per_second": 1.026,
       "train_steps_per_second": 0.032
     }
   ],
   "logging_steps": 5,
-  "max_steps": 229,
+  "max_steps": 234,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
@@ -365,7 +372,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 47895596236800.0,
+  "total_flos": 48942494515200.0,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7e3bd716872e859fc26cb2d83509c29b1c0e5972c73d760b0ac429e559979835
+oid sha256:53ce655e7ab1129bcd883d7974bca1f5c5778062e5c8a4eb41ef38796a8ee647
 size 6520