diff --git a/adapter/adapter_config.json b/adapter/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/adapter_config.json +++ b/adapter/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/adapter_model.safetensors b/adapter/adapter_model.safetensors index 861fbb9f85ceef292d1e10aec3f3d60b2fe00d9a..b7733ff441ada9ccc544a5d5bf1c6a52935bd15e 100644 --- a/adapter/adapter_model.safetensors +++ b/adapter/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:98c1a96e3957d6f2d8ac76a92aaead06fd6271985eba9a3c6751e465f80dd2af -size 59934640 +oid sha256:e3be2c8518433696a6aa5b088a43823facc099b20c7beab385e9352ec3023f2e +size 239536776 diff --git a/adapter/checkpoint-1/adapter_config.json b/adapter/checkpoint-1/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-1/adapter_config.json +++ b/adapter/checkpoint-1/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-1/adapter_model.safetensors b/adapter/checkpoint-1/adapter_model.safetensors index 5462b22359b25bc2d25017f09f12c28bc1e848a8..f38e3be07406e259ce92afab3b4715990e026ef1 100644 --- a/adapter/checkpoint-1/adapter_model.safetensors +++ b/adapter/checkpoint-1/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:21ce0fa1f789d5239dc91e9e1f9389a8890548841e35e2991a117a4268853a82 -size 59934640 +oid sha256:70c3c9e261473ae17f27314b2bba1b03909921856bf5c133d5e2078fe6f26477 +size 239536776 diff --git a/adapter/checkpoint-1/optimizer.pt b/adapter/checkpoint-1/optimizer.pt index 5b361e2b9050411114c83cf02a00a8849b34b068..aef7e93670686ecfbbee9916d0303692dac174a9 100644 --- a/adapter/checkpoint-1/optimizer.pt +++ b/adapter/checkpoint-1/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:176f12959d3e8db149fbdc04eb8859cadc5107cc37a4d7e4268df2bf83774c5c -size 61403589 +oid sha256:9e085fd55cad63de4fbae77072082b27df5ccf6b6498e66ab3b053bc24aa5a34 +size 243818437 diff --git a/adapter/checkpoint-1/trainer_state.json b/adapter/checkpoint-1/trainer_state.json index 6b5239d0477b51d4360e23e82fafca20e6f6e95e..13c3c52b8dc2dc3c648a894d150e6ca4d2d891ff 100644 --- a/adapter/checkpoint-1/trainer_state.json +++ b/adapter/checkpoint-1/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 12301446340608.0, + "total_flos": 12695312228352.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-10/adapter_config.json b/adapter/checkpoint-10/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-10/adapter_config.json +++ b/adapter/checkpoint-10/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-10/adapter_model.safetensors b/adapter/checkpoint-10/adapter_model.safetensors index e4f0f5265c6905d438d72b7f60c77e9271d8bb33..ffb7514976a1c3b6b066f127037ee57121ebb387 100644 --- a/adapter/checkpoint-10/adapter_model.safetensors +++ b/adapter/checkpoint-10/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:a29e0d2a70dfb944c3f4843d60865ff9804ad190500e952e50f8d17b8b71d14b -size 59934640 +oid sha256:05d17ae14cc83cb1bacf66ddbf8a731aa95e7ee88c27331d712b7f05316ab93b +size 239536776 diff --git a/adapter/checkpoint-10/optimizer.pt b/adapter/checkpoint-10/optimizer.pt index efa0cd7b63c46e1e0db9d183119e18629fe5d298..5bee592948320dba1f104883751d009efbba2193 100644 --- a/adapter/checkpoint-10/optimizer.pt +++ b/adapter/checkpoint-10/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:b70b2be656da4eaa67e78914f91447b6c2cb786487d5cc3550ca0b4a81830fa1 -size 61403589 +oid sha256:5f101ceedd36da51a74e912ba21ac9744d4bcf49d3acf3677c9a6a4f2f45e9b5 +size 243818437 diff --git a/adapter/checkpoint-10/trainer_state.json b/adapter/checkpoint-10/trainer_state.json index 67e667760c8ee65adfa30df3e067d2ecd2bf9447..edbb0fdfa9e8d4554f74928d0aae3ef0dd229230 100644 --- a/adapter/checkpoint-10/trainer_state.json +++ b/adapter/checkpoint-10/trainer_state.json @@ -10,12 +10,12 @@ "is_world_process_zero": true, "log_history": [ { - "entropy": 1.4535226494073867, + "entropy": 0.8513693287968636, "epoch": 10.0, - "grad_norm": 3.046875, + "grad_norm": 1.34375, "learning_rate": 8e-05, - "loss": 2.295806121826172, - "mean_token_accuracy": 0.6982592329382896, + "loss": 1.310697078704834, + "mean_token_accuracy": 0.8319162212312221, "num_tokens": 6910.0, "step": 10 } @@ -37,7 +37,7 @@ "attributes": {} } }, - "total_flos": 124394379411456.0, + "total_flos": 128377220235264.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-11/adapter_config.json b/adapter/checkpoint-11/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-11/adapter_config.json +++ b/adapter/checkpoint-11/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-11/adapter_model.safetensors b/adapter/checkpoint-11/adapter_model.safetensors index 63ec605fc09ce555e0bc114b1450c00e0ad88b8c..5db8a313ccfab87261d2d0bea6c5bbc6d411bc92 100644 --- a/adapter/checkpoint-11/adapter_model.safetensors +++ b/adapter/checkpoint-11/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:aee871bd99a9f7b687207d436cb48bb2e99ce3fe4a49d0aa9f2b5adf321bdc72 -size 59934640 +oid sha256:9d9681efd88617c5414ad8a2f16a9fe33ff44ea2cff0324cee219e52ebabff7c +size 239536776 diff --git a/adapter/checkpoint-11/optimizer.pt b/adapter/checkpoint-11/optimizer.pt index c3ca3654b7597373cc1f8ba8df9762995c5f0f4b..5eefaafba2fcd9dd087d2e950fe46ba389c196cb 100644 --- a/adapter/checkpoint-11/optimizer.pt +++ b/adapter/checkpoint-11/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:68845f446c5dde79cd8ec515256965cae52629ec6ae88c266df936570d3a26fd -size 61403589 +oid sha256:2fa7a3a67810ffd51f4413bb310c5e0b07749a8aef580fb8d5694adec4318e1a +size 243818437 diff --git a/adapter/checkpoint-11/trainer_state.json b/adapter/checkpoint-11/trainer_state.json index 24e25e9d01fdf45bdb7dd679aad1949093374895..36982e121c080883b30cf9898bffa33ffd33651e 100644 --- a/adapter/checkpoint-11/trainer_state.json +++ b/adapter/checkpoint-11/trainer_state.json @@ -10,12 +10,12 @@ "is_world_process_zero": true, "log_history": [ { - "entropy": 1.4535226494073867, + "entropy": 0.8513693287968636, "epoch": 10.0, - "grad_norm": 3.046875, + "grad_norm": 1.34375, "learning_rate": 8e-05, - "loss": 2.295806121826172, - "mean_token_accuracy": 0.6982592329382896, + "loss": 1.310697078704834, + "mean_token_accuracy": 0.8319162212312221, "num_tokens": 6910.0, "step": 10 } @@ -37,7 +37,7 @@ "attributes": {} } }, - "total_flos": 137082875363328.0, + "total_flos": 141471974572032.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-12/adapter_config.json b/adapter/checkpoint-12/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-12/adapter_config.json +++ b/adapter/checkpoint-12/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-12/adapter_model.safetensors b/adapter/checkpoint-12/adapter_model.safetensors index 5fc139c13b8a84732a93e57eaa24cc144456a8d8..86119b0cd91512a2de49d44455878a185e5c3d7c 100644 --- a/adapter/checkpoint-12/adapter_model.safetensors +++ b/adapter/checkpoint-12/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:fe54aa525c7578a6b9fccf706a68ba1efeb01af91468da9dfc33e63809648845 -size 59934640 +oid sha256:6e2df4368fe55f5a79123103fc8d9321193f5a1c9b4771867f8c8a0e1d214108 +size 239536776 diff --git a/adapter/checkpoint-12/optimizer.pt b/adapter/checkpoint-12/optimizer.pt index 7086904eb5a8f09ba340d46b984e0a676ab4ae19..42fdf661ed6a37c6dea4dccdec553095eab890c7 100644 --- a/adapter/checkpoint-12/optimizer.pt +++ b/adapter/checkpoint-12/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:87cff6786c1a5d91bbc130aef6c8654bed7dfc381eb68d75ea035bccbaf93034 -size 61403589 +oid sha256:effe09760ff16f3e0e212c42ab9a3bf3df4644871b0cd48be8e746b423d3045a +size 243818437 diff --git a/adapter/checkpoint-12/trainer_state.json b/adapter/checkpoint-12/trainer_state.json index 9cca076840fabd66b11557c73509f087b6533a10..0a146af766721317b8649f537f0d22ffb7495671 100644 --- a/adapter/checkpoint-12/trainer_state.json +++ b/adapter/checkpoint-12/trainer_state.json @@ -10,12 +10,12 @@ "is_world_process_zero": true, "log_history": [ { - "entropy": 1.4535226494073867, + "entropy": 0.8513693287968636, "epoch": 10.0, - "grad_norm": 3.046875, + "grad_norm": 1.34375, "learning_rate": 8e-05, - "loss": 2.295806121826172, - "mean_token_accuracy": 0.6982592329382896, + "loss": 1.310697078704834, + "mean_token_accuracy": 0.8319162212312221, "num_tokens": 6910.0, "step": 10 } @@ -37,7 +37,7 @@ "attributes": {} } }, - "total_flos": 149653573607424.0, + "total_flos": 154445159571456.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-13/adapter_config.json b/adapter/checkpoint-13/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-13/adapter_config.json +++ b/adapter/checkpoint-13/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-13/adapter_model.safetensors b/adapter/checkpoint-13/adapter_model.safetensors index 70f0bd269fe7e5b6af7e7661d1789d49672e0e38..03ae81dfd6e2a3239cc66bfee6ff78ac4718bc3a 100644 --- a/adapter/checkpoint-13/adapter_model.safetensors +++ b/adapter/checkpoint-13/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:45b40737c9ddc72ae19e833e08e169e5f1002bbd9aff96ce344550b4db23d7e5 -size 59934640 +oid sha256:90c8bf5cc5e3a9080536acd4e22d9b395f3b96ac285a926ec195764e9a3ac920 +size 239536776 diff --git a/adapter/checkpoint-13/optimizer.pt b/adapter/checkpoint-13/optimizer.pt index aead7d91de007b9e542cdce76dce8b78f61248ef..7cc75c0cfdf64170762c60f3e4563b2f267b352b 100644 --- a/adapter/checkpoint-13/optimizer.pt +++ b/adapter/checkpoint-13/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:bdfa6deb4f989862f3cd407e10f2ac9d06d1137109a053701d5ce06c6a68e617 -size 61403589 +oid sha256:8de3a29fd07f2ee3595d0ac98721df096041694f03ff5aa11649d155f5ecb5ca +size 243818437 diff --git a/adapter/checkpoint-13/trainer_state.json b/adapter/checkpoint-13/trainer_state.json index e288c088d3d1289acffe548537ae4f35ea420958..647e953f449c6cdc0ab1df8e98363255e8ce43c7 100644 --- a/adapter/checkpoint-13/trainer_state.json +++ b/adapter/checkpoint-13/trainer_state.json @@ -10,12 +10,12 @@ "is_world_process_zero": true, "log_history": [ { - "entropy": 1.4535226494073867, + "entropy": 0.8513693287968636, "epoch": 10.0, - "grad_norm": 3.046875, + "grad_norm": 1.34375, "learning_rate": 8e-05, - "loss": 2.295806121826172, - "mean_token_accuracy": 0.6982592329382896, + "loss": 1.310697078704834, + "mean_token_accuracy": 0.8319162212312221, "num_tokens": 6910.0, "step": 10 } @@ -37,7 +37,7 @@ "attributes": {} } }, - "total_flos": 162072817655808.0, + "total_flos": 167262041137152.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-14/adapter_config.json b/adapter/checkpoint-14/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-14/adapter_config.json +++ b/adapter/checkpoint-14/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-14/adapter_model.safetensors b/adapter/checkpoint-14/adapter_model.safetensors index eb5666ba0422c9a697811916002aa03b9f999039..cdac2c0ab380df7093481859bcc49553623437ed 100644 --- a/adapter/checkpoint-14/adapter_model.safetensors +++ b/adapter/checkpoint-14/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:a476e70c9e67048c4892fd9f5e9f70ded338d85ef5c44a86141c2ce49763a1ee -size 59934640 +oid sha256:d50aa8b388ec8733ac44bd472b9465ce8dd25750a07b47bb921b8abe78e93449 +size 239536776 diff --git a/adapter/checkpoint-14/optimizer.pt b/adapter/checkpoint-14/optimizer.pt index cb3b8f999edc31354ba772a7ebac7a707a925730..25b100b2e210c879ed81b0800681353cf5e73aff 100644 --- a/adapter/checkpoint-14/optimizer.pt +++ b/adapter/checkpoint-14/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:b11d2a906c1b8dda251b88971127a4508d57b0e527c64be76bc3ce5c13b72406 -size 61403589 +oid sha256:b435012f854d460c7c66d1e22a7d951f7b7fc22aa9f957bc2f757111325d4c16 +size 243818437 diff --git a/adapter/checkpoint-14/trainer_state.json b/adapter/checkpoint-14/trainer_state.json index 4544627ee9034ae75f7a43ad46034ce518b2c072..b2edd6ca4191a487abd2fb2eccf5976fc6101de8 100644 --- a/adapter/checkpoint-14/trainer_state.json +++ b/adapter/checkpoint-14/trainer_state.json @@ -10,12 +10,12 @@ "is_world_process_zero": true, "log_history": [ { - "entropy": 1.4535226494073867, + "entropy": 0.8513693287968636, "epoch": 10.0, - "grad_norm": 3.046875, + "grad_norm": 1.34375, "learning_rate": 8e-05, - "loss": 2.295806121826172, - "mean_token_accuracy": 0.6982592329382896, + "loss": 1.310697078704834, + "mean_token_accuracy": 0.8319162212312221, "num_tokens": 6910.0, "step": 10 } @@ -37,7 +37,7 @@ "attributes": {} } }, - "total_flos": 174643515899904.0, + "total_flos": 180235226136576.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-15/adapter_config.json b/adapter/checkpoint-15/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-15/adapter_config.json +++ b/adapter/checkpoint-15/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-15/adapter_model.safetensors b/adapter/checkpoint-15/adapter_model.safetensors index 861fbb9f85ceef292d1e10aec3f3d60b2fe00d9a..b7733ff441ada9ccc544a5d5bf1c6a52935bd15e 100644 --- a/adapter/checkpoint-15/adapter_model.safetensors +++ b/adapter/checkpoint-15/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:98c1a96e3957d6f2d8ac76a92aaead06fd6271985eba9a3c6751e465f80dd2af -size 59934640 +oid sha256:e3be2c8518433696a6aa5b088a43823facc099b20c7beab385e9352ec3023f2e +size 239536776 diff --git a/adapter/checkpoint-15/optimizer.pt b/adapter/checkpoint-15/optimizer.pt index b92b2197c4b6785b7c6d75ff32398da3fbdddb1d..df78b83007f32859a84d45b7a31485ec1148bcb5 100644 --- a/adapter/checkpoint-15/optimizer.pt +++ b/adapter/checkpoint-15/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:7c7b95523d0eb0840773e7a8a89b454480020e49a49280c0ee0d138341c3598f -size 61403589 +oid sha256:b4935ec169e25391ef800c9705cd37a3a5b35a3c6cde2dd9395cae3ff03c3ee2 +size 243818437 diff --git a/adapter/checkpoint-15/trainer_state.json b/adapter/checkpoint-15/trainer_state.json index 7d8ea5b08ce1cc2e28a2e508d066981c18540d4f..f84b0825d292ddace1a5c055a2db3dcaa203d2b0 100644 --- a/adapter/checkpoint-15/trainer_state.json +++ b/adapter/checkpoint-15/trainer_state.json @@ -10,12 +10,12 @@ "is_world_process_zero": true, "log_history": [ { - "entropy": 1.4535226494073867, + "entropy": 0.8513693287968636, "epoch": 10.0, - "grad_norm": 3.046875, + "grad_norm": 1.34375, "learning_rate": 8e-05, - "loss": 2.295806121826172, - "mean_token_accuracy": 0.6982592329382896, + "loss": 1.310697078704834, + "mean_token_accuracy": 0.8319162212312221, "num_tokens": 6910.0, "step": 10 } @@ -37,7 +37,7 @@ "attributes": {} } }, - "total_flos": 187332011851776.0, + "total_flos": 193329980473344.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-2/adapter_config.json b/adapter/checkpoint-2/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-2/adapter_config.json +++ b/adapter/checkpoint-2/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-2/adapter_model.safetensors b/adapter/checkpoint-2/adapter_model.safetensors index a93d65d272b4d3a19d96452efe3194bcd3c98c2d..105f732bb64e19ddf04aafad5561a8bf6195e6ba 100644 --- a/adapter/checkpoint-2/adapter_model.safetensors +++ b/adapter/checkpoint-2/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:9e59d9845c2883cb094ca32fd920c1c51e66ddd69e86c6b02fd46e3f3051cddc -size 59934640 +oid sha256:43e7a048bea5c9b88eae829f5a6ea192d63a9a9ee3be127d8931054344774a0b +size 239536776 diff --git a/adapter/checkpoint-2/optimizer.pt b/adapter/checkpoint-2/optimizer.pt index a42a247f3791d9fc14b31eedc8d6bee5d9e71eab..914df82d45ded461ae784a61f1e9fd2c9f7f83c2 100644 --- a/adapter/checkpoint-2/optimizer.pt +++ b/adapter/checkpoint-2/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:6d0bc9dc9587e31a2378b3bc737cb6de7d1e512c821ec342a7b5c4d24db76862 -size 61403589 +oid sha256:b7b530d3378e9cbb1e3f4ffbcd10e7d058234260cffaeab63775acb44a4da560 +size 243818437 diff --git a/adapter/checkpoint-2/trainer_state.json b/adapter/checkpoint-2/trainer_state.json index 3d37b785e38542d20dd4d0027f7145e6f84d6b6b..c4ed5f3caaa249816c13c642dae2ee82d3f81595 100644 --- a/adapter/checkpoint-2/trainer_state.json +++ b/adapter/checkpoint-2/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 24922629316608.0, + "total_flos": 25720598372352.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-3/adapter_config.json b/adapter/checkpoint-3/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-3/adapter_config.json +++ b/adapter/checkpoint-3/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-3/adapter_model.safetensors b/adapter/checkpoint-3/adapter_model.safetensors index b684cc124505a457d05e06b9a3e071062377e8f5..e855914feea142cee6308e63e925ac666d96f870 100644 --- a/adapter/checkpoint-3/adapter_model.safetensors +++ b/adapter/checkpoint-3/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:1271a02ac30b811df23bf6d4ded5597b51c806f27458fc63f2790f6be6ef98f0 -size 59934640 +oid sha256:63a2df1a70456844c689f40525f9c3ac54f76c0275316566a7b60d88668ba86d +size 239536776 diff --git a/adapter/checkpoint-3/optimizer.pt b/adapter/checkpoint-3/optimizer.pt index f399a95aad0d7d5858f8b160f873d86c1eda8fbb..6e8c94f8b03681f5b7f9962e28152e02a4b504d0 100644 --- a/adapter/checkpoint-3/optimizer.pt +++ b/adapter/checkpoint-3/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:8909fc0348e069e67578ae1ecd1dfc2ce7ab9dc95ae0178d58c4a4aa7c853417 -size 61403589 +oid sha256:ea813c605e804352ba8ff66dd96e9f8683da294ca935813cb507520dff866fd7 +size 243818437 diff --git a/adapter/checkpoint-3/trainer_state.json b/adapter/checkpoint-3/trainer_state.json index b95390fde62b05572b2e242256bcc2afd7a197e8..e6bd487e8644a683d67df043d6445f6c9a13c2a7 100644 --- a/adapter/checkpoint-3/trainer_state.json +++ b/adapter/checkpoint-3/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 37106277949440.0, + "total_flos": 38294341263360.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-4/adapter_config.json b/adapter/checkpoint-4/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-4/adapter_config.json +++ b/adapter/checkpoint-4/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-4/adapter_model.safetensors b/adapter/checkpoint-4/adapter_model.safetensors index dfe22174db1537962bc94692369516d89ab6f8c3..c80c673bb4f6e042d5430db2350652a37252090e 100644 --- a/adapter/checkpoint-4/adapter_model.safetensors +++ b/adapter/checkpoint-4/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:495fd96b71521f51b0859ca6c66f732ce4795d2860d8ecb74eb7d91cbfe53675 -size 59934640 +oid sha256:0b759651321e274d2fac4ede5eae09ec051c644ddc3da7cd711b4fee072ff7ab +size 239536776 diff --git a/adapter/checkpoint-4/optimizer.pt b/adapter/checkpoint-4/optimizer.pt index 5ee305b97749348dfb578586697d240050cf673d..49d5f9ed7ebb2d2af90cf782c9ba97b429e28ab9 100644 --- a/adapter/checkpoint-4/optimizer.pt +++ b/adapter/checkpoint-4/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:bd5f2611f0f07200fcffd7e94468ef7c059fcbac076a3b4d287c130d8fb82f7a -size 61403589 +oid sha256:d70fbb46583d0982619b54aa0647e44b07513ed9d5bbe96c792d690c4cd82385 +size 243818437 diff --git a/adapter/checkpoint-4/trainer_state.json b/adapter/checkpoint-4/trainer_state.json index 8d78246851caa7f5d4b1cf5e5083b0a77387d11c..fad0fab9cee057d34c1fe9de3a5dec16e360dec3 100644 --- a/adapter/checkpoint-4/trainer_state.json +++ b/adapter/checkpoint-4/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 49542350241792.0, + "total_flos": 51128589877248.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-5/adapter_config.json b/adapter/checkpoint-5/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-5/adapter_config.json +++ b/adapter/checkpoint-5/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-5/adapter_model.safetensors b/adapter/checkpoint-5/adapter_model.safetensors index 45493b3d785244ea435c22a3a735c3921457a176..d41707fb248b149938043e3ab88bbc556ebe9ba6 100644 --- a/adapter/checkpoint-5/adapter_model.safetensors +++ b/adapter/checkpoint-5/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:a33586140579f464c809086aced36403d8849cd2c524bed00d56a3271d3c3dab -size 59934640 +oid sha256:29e56d014bad376378d3b34998bcb3b3b9fb51b51b3f5db99fc3198c9852018d +size 239536776 diff --git a/adapter/checkpoint-5/optimizer.pt b/adapter/checkpoint-5/optimizer.pt index 6a6449353213185b8fc5db358aa449ff4a6692cf..75b68afe0460ed295b133d5316d84c5eca14dc8c 100644 --- a/adapter/checkpoint-5/optimizer.pt +++ b/adapter/checkpoint-5/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:684e94bfb97cd50c3d8cbe491de291ceafd499b7fa01a3f05faaa92a625b13ea -size 61403589 +oid sha256:c01e814587fc44eb23dc47e7d9794565fa2c4a0982fc0e8ba8e92e6933f5bf3e +size 243818437 diff --git a/adapter/checkpoint-5/trainer_state.json b/adapter/checkpoint-5/trainer_state.json index 009671cc199abbf53234811021e3b8267d9eb5bf..aa3f60f93b55132ee8e74ebf6f16ff8a71baa622 100644 --- a/adapter/checkpoint-5/trainer_state.json +++ b/adapter/checkpoint-5/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 61978422534144.0, + "total_flos": 63962838491136.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-6/adapter_config.json b/adapter/checkpoint-6/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-6/adapter_config.json +++ b/adapter/checkpoint-6/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-6/adapter_model.safetensors b/adapter/checkpoint-6/adapter_model.safetensors index 182a22d377a7ca884acf8f209dd0b30c44b651f3..d2f65c2b6f544b00910b8e7358a5aa3407e58d40 100644 --- a/adapter/checkpoint-6/adapter_model.safetensors +++ b/adapter/checkpoint-6/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:f392c3367758e68ae2a6335f7273eebcc85a9a468fab484f4c404fa526a62f1b -size 59934640 +oid sha256:2d520e27667af63bf65784a8245076234ec27b98f7e084552481b2c0ac118a13 +size 239536776 diff --git a/adapter/checkpoint-6/optimizer.pt b/adapter/checkpoint-6/optimizer.pt index ae3b83146db90cc247ef2484ae01a62c48accde5..e1b910fe4cd7b2897404ccd9e70d1dfe106fe89d 100644 --- a/adapter/checkpoint-6/optimizer.pt +++ b/adapter/checkpoint-6/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:5e5f490cfc86381235bdeb3c91a427d42571452be978214b1f1ec171bd223ff8 -size 61403589 +oid sha256:2fa67e6c425f93c8b24ac0d40e17dd9d2e7902edc418a3ff7636da87ffc1da8f +size 243818437 diff --git a/adapter/checkpoint-6/trainer_state.json b/adapter/checkpoint-6/trainer_state.json index 83456c63db3f86cf8d4376670c54e899b70ecccd..40f40f1b197d4094a8410f289276031f9c8214b4 100644 --- a/adapter/checkpoint-6/trainer_state.json +++ b/adapter/checkpoint-6/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 74532292534272.0, + "total_flos": 76918656442368.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-7/adapter_config.json b/adapter/checkpoint-7/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-7/adapter_config.json +++ b/adapter/checkpoint-7/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-7/adapter_model.safetensors b/adapter/checkpoint-7/adapter_model.safetensors index f6fad09c3cda14dc88513e9fbe5b6d808af9d114..2657da0dee3a433805a41b486c2626d9539cfede 100644 --- a/adapter/checkpoint-7/adapter_model.safetensors +++ b/adapter/checkpoint-7/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:4800f0a0b5568c53fa2e3949c1719587339c8f6d684b3aa1e92b04dc4beb1869 -size 59934640 +oid sha256:6c8b39faf3a2c769ccbcd0f4bf97596484347460e0b96cdc227c636c8f7e523d +size 239536776 diff --git a/adapter/checkpoint-7/optimizer.pt b/adapter/checkpoint-7/optimizer.pt index a187fd61608bb97ceafea50e4a992f766a0370c9..189422e47e67b55620fbb03ae95db5926ef1cceb 100644 --- a/adapter/checkpoint-7/optimizer.pt +++ b/adapter/checkpoint-7/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:b3b0cd4e3cd6cd0c9cc4832d34132b6a57da0a4e1b044d85cb04725c21156d61 -size 61403589 +oid sha256:b94dbe7064b764155bb9e2b3d791bb4f4a5fa064cd4d2ffbbd4fbb5fdcd4b70b +size 243818437 diff --git a/adapter/checkpoint-7/trainer_state.json b/adapter/checkpoint-7/trainer_state.json index f9acb87d42989c0eed07a66b65ca830cc10a89ee..f32f3365f69ab6ffe6049ffa9a3688804b5ec51c 100644 --- a/adapter/checkpoint-7/trainer_state.json +++ b/adapter/checkpoint-7/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 86884223606784.0, + "total_flos": 89666069815296.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-8/adapter_config.json b/adapter/checkpoint-8/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-8/adapter_config.json +++ b/adapter/checkpoint-8/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-8/adapter_model.safetensors b/adapter/checkpoint-8/adapter_model.safetensors index 809995a3a84a4cff71f0fe215d43c666654c97d5..b7aab915274c1413ff4c2f16d13986057f3a438a 100644 --- a/adapter/checkpoint-8/adapter_model.safetensors +++ b/adapter/checkpoint-8/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:65b9c96a063aa561b3ea5699c4c93cea1b580bd69a2358c2f3a20e00e80256b5 -size 59934640 +oid sha256:00f650b9434b1e9710dbda8ecdf0913aabff56e76246b504a909c8bf1cfd5820 +size 239536776 diff --git a/adapter/checkpoint-8/optimizer.pt b/adapter/checkpoint-8/optimizer.pt index c89661bfc74b853e3a0526e0635206664b740a6f..2fdaab061a6b40bbbe1fab0d8c4d4b6fac5e01ab 100644 --- a/adapter/checkpoint-8/optimizer.pt +++ b/adapter/checkpoint-8/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:70caad6901b3c8339bc78c7173553409448d2199371b064cfbe7539c028105a8 -size 61403589 +oid sha256:16011f2abf7e734e798897529fdb53e299bc3ae0e164decc6417556b83bc625f +size 243818437 diff --git a/adapter/checkpoint-8/trainer_state.json b/adapter/checkpoint-8/trainer_state.json index a4d3e5f12e3a372c2a38698826264aa5005ae55e..e641fa638bb24ddca6b53b015ada91a6ff94ba26 100644 --- a/adapter/checkpoint-8/trainer_state.json +++ b/adapter/checkpoint-8/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 99252982923264.0, + "total_flos": 102430850236416.0, "train_batch_size": 4, "trial_name": null, "trial_params": null diff --git a/adapter/checkpoint-9/adapter_config.json b/adapter/checkpoint-9/adapter_config.json index a94aa479415cb712a22e6dac8d3f23d6ed08addb..e934c4f8a7385575225a5b73e1281e2a8fbdb3e3 100644 --- a/adapter/checkpoint-9/adapter_config.json +++ b/adapter/checkpoint-9/adapter_config.json @@ -16,7 +16,7 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 32, + "lora_alpha": 256, "lora_bias": false, "lora_dropout": 0.05, "lora_ga_config": null, @@ -27,16 +27,16 @@ "peft_type": "LORA", "peft_version": "0.20.0", "qalora_group_size": 16, - "r": 16, + "r": 64, "rank_pattern": {}, "revision": null, "target_modules": [ - "gate_proj", - "up_proj", "o_proj", - "v_proj", - "down_proj", + "up_proj", "k_proj", + "gate_proj", + "down_proj", + "v_proj", "q_proj" ], "target_parameters": null, diff --git a/adapter/checkpoint-9/adapter_model.safetensors b/adapter/checkpoint-9/adapter_model.safetensors index 94fb8c519aa0f669b1a186f397a24cd1ba1fe928..c00dbef482dc9b2a443074747d745a416c0c1b78 100644 --- a/adapter/checkpoint-9/adapter_model.safetensors +++ b/adapter/checkpoint-9/adapter_model.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:ad239db9dad57a78f93a8d316f568992e6c44bb7f63bb731f270880935010b1b -size 59934640 +oid sha256:034d5337cafb6b34232eb6a40723e4d87caba75cdc73b10f25fa97eefb8a99a5 +size 239536776 diff --git a/adapter/checkpoint-9/optimizer.pt b/adapter/checkpoint-9/optimizer.pt index b34a36af1963882c8d723cad2633e1addd872462..386f8c313d409fbd535b165a747332a2488c67c7 100644 --- a/adapter/checkpoint-9/optimizer.pt +++ b/adapter/checkpoint-9/optimizer.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:91b2d4bc65d0633da540784515725d9d7dbabfa860959c72d2acccd88ce7c611 -size 61403589 +oid sha256:7567a4fd6835febeb5724bd7394d79ac797cefd8b7d72ab41dc830ecf9b2e5e5 +size 243818437 diff --git a/adapter/checkpoint-9/trainer_state.json b/adapter/checkpoint-9/trainer_state.json index c39155ffaf06476d3807af2ca296e04be4acd451..c85afb23a64ad0723f8178206a8a2b70f7a22025 100644 --- a/adapter/checkpoint-9/trainer_state.json +++ b/adapter/checkpoint-9/trainer_state.json @@ -26,7 +26,7 @@ "attributes": {} } }, - "total_flos": 111958307119104.0, + "total_flos": 115542971621376.0, "train_batch_size": 4, "trial_name": null, "trial_params": null