diff --git a/gpt256.bsh b/gpt256.bsh new file mode 100644 index 0000000000000000000000000000000000000000..6ebbe5feb6d4bcdceaaf60d1f974a4f556fc82b4 --- /dev/null +++ b/gpt256.bsh @@ -0,0 +1,349 @@ +# This is a base shape file encoded in yaml +# - `null` indicates a dimension is "finite", i.e. a non-"width" dimension +# - a number indicates the base dimension of an "infinite" dimension, i.e. some notion of "width" +transformer.h.0.attn.c_attn.bias: +- 384 +transformer.h.0.attn.c_attn.weight: +- 128 +- 384 +transformer.h.0.attn.c_proj.bias: +- 128 +transformer.h.0.attn.c_proj.weight: +- 128 +- 128 +transformer.h.0.ln_1.bias: +- 128 +transformer.h.0.ln_1.weight: +- 128 +transformer.h.0.ln_2.bias: +- 128 +transformer.h.0.ln_2.weight: +- 128 +transformer.h.0.mlp.c_fc.bias: +- 512 +transformer.h.0.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.0.mlp.c_proj.bias: +- 128 +transformer.h.0.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.1.attn.c_attn.bias: +- 384 +transformer.h.1.attn.c_attn.weight: +- 128 +- 384 +transformer.h.1.attn.c_proj.bias: +- 128 +transformer.h.1.attn.c_proj.weight: +- 128 +- 128 +transformer.h.1.ln_1.bias: +- 128 +transformer.h.1.ln_1.weight: +- 128 +transformer.h.1.ln_2.bias: +- 128 +transformer.h.1.ln_2.weight: +- 128 +transformer.h.1.mlp.c_fc.bias: +- 512 +transformer.h.1.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.1.mlp.c_proj.bias: +- 128 +transformer.h.1.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.10.attn.c_attn.bias: +- 384 +transformer.h.10.attn.c_attn.weight: +- 128 +- 384 +transformer.h.10.attn.c_proj.bias: +- 128 +transformer.h.10.attn.c_proj.weight: +- 128 +- 128 +transformer.h.10.ln_1.bias: +- 128 +transformer.h.10.ln_1.weight: +- 128 +transformer.h.10.ln_2.bias: +- 128 +transformer.h.10.ln_2.weight: +- 128 +transformer.h.10.mlp.c_fc.bias: +- 512 +transformer.h.10.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.10.mlp.c_proj.bias: +- 128 +transformer.h.10.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.11.attn.c_attn.bias: +- 384 +transformer.h.11.attn.c_attn.weight: +- 128 +- 384 +transformer.h.11.attn.c_proj.bias: +- 128 +transformer.h.11.attn.c_proj.weight: +- 128 +- 128 +transformer.h.11.ln_1.bias: +- 128 +transformer.h.11.ln_1.weight: +- 128 +transformer.h.11.ln_2.bias: +- 128 +transformer.h.11.ln_2.weight: +- 128 +transformer.h.11.mlp.c_fc.bias: +- 512 +transformer.h.11.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.11.mlp.c_proj.bias: +- 128 +transformer.h.11.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.2.attn.c_attn.bias: +- 384 +transformer.h.2.attn.c_attn.weight: +- 128 +- 384 +transformer.h.2.attn.c_proj.bias: +- 128 +transformer.h.2.attn.c_proj.weight: +- 128 +- 128 +transformer.h.2.ln_1.bias: +- 128 +transformer.h.2.ln_1.weight: +- 128 +transformer.h.2.ln_2.bias: +- 128 +transformer.h.2.ln_2.weight: +- 128 +transformer.h.2.mlp.c_fc.bias: +- 512 +transformer.h.2.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.2.mlp.c_proj.bias: +- 128 +transformer.h.2.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.3.attn.c_attn.bias: +- 384 +transformer.h.3.attn.c_attn.weight: +- 128 +- 384 +transformer.h.3.attn.c_proj.bias: +- 128 +transformer.h.3.attn.c_proj.weight: +- 128 +- 128 +transformer.h.3.ln_1.bias: +- 128 +transformer.h.3.ln_1.weight: +- 128 +transformer.h.3.ln_2.bias: +- 128 +transformer.h.3.ln_2.weight: +- 128 +transformer.h.3.mlp.c_fc.bias: +- 512 +transformer.h.3.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.3.mlp.c_proj.bias: +- 128 +transformer.h.3.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.4.attn.c_attn.bias: +- 384 +transformer.h.4.attn.c_attn.weight: +- 128 +- 384 +transformer.h.4.attn.c_proj.bias: +- 128 +transformer.h.4.attn.c_proj.weight: +- 128 +- 128 +transformer.h.4.ln_1.bias: +- 128 +transformer.h.4.ln_1.weight: +- 128 +transformer.h.4.ln_2.bias: +- 128 +transformer.h.4.ln_2.weight: +- 128 +transformer.h.4.mlp.c_fc.bias: +- 512 +transformer.h.4.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.4.mlp.c_proj.bias: +- 128 +transformer.h.4.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.5.attn.c_attn.bias: +- 384 +transformer.h.5.attn.c_attn.weight: +- 128 +- 384 +transformer.h.5.attn.c_proj.bias: +- 128 +transformer.h.5.attn.c_proj.weight: +- 128 +- 128 +transformer.h.5.ln_1.bias: +- 128 +transformer.h.5.ln_1.weight: +- 128 +transformer.h.5.ln_2.bias: +- 128 +transformer.h.5.ln_2.weight: +- 128 +transformer.h.5.mlp.c_fc.bias: +- 512 +transformer.h.5.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.5.mlp.c_proj.bias: +- 128 +transformer.h.5.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.6.attn.c_attn.bias: +- 384 +transformer.h.6.attn.c_attn.weight: +- 128 +- 384 +transformer.h.6.attn.c_proj.bias: +- 128 +transformer.h.6.attn.c_proj.weight: +- 128 +- 128 +transformer.h.6.ln_1.bias: +- 128 +transformer.h.6.ln_1.weight: +- 128 +transformer.h.6.ln_2.bias: +- 128 +transformer.h.6.ln_2.weight: +- 128 +transformer.h.6.mlp.c_fc.bias: +- 512 +transformer.h.6.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.6.mlp.c_proj.bias: +- 128 +transformer.h.6.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.7.attn.c_attn.bias: +- 384 +transformer.h.7.attn.c_attn.weight: +- 128 +- 384 +transformer.h.7.attn.c_proj.bias: +- 128 +transformer.h.7.attn.c_proj.weight: +- 128 +- 128 +transformer.h.7.ln_1.bias: +- 128 +transformer.h.7.ln_1.weight: +- 128 +transformer.h.7.ln_2.bias: +- 128 +transformer.h.7.ln_2.weight: +- 128 +transformer.h.7.mlp.c_fc.bias: +- 512 +transformer.h.7.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.7.mlp.c_proj.bias: +- 128 +transformer.h.7.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.8.attn.c_attn.bias: +- 384 +transformer.h.8.attn.c_attn.weight: +- 128 +- 384 +transformer.h.8.attn.c_proj.bias: +- 128 +transformer.h.8.attn.c_proj.weight: +- 128 +- 128 +transformer.h.8.ln_1.bias: +- 128 +transformer.h.8.ln_1.weight: +- 128 +transformer.h.8.ln_2.bias: +- 128 +transformer.h.8.ln_2.weight: +- 128 +transformer.h.8.mlp.c_fc.bias: +- 512 +transformer.h.8.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.8.mlp.c_proj.bias: +- 128 +transformer.h.8.mlp.c_proj.weight: +- 512 +- 128 +transformer.h.9.attn.c_attn.bias: +- 384 +transformer.h.9.attn.c_attn.weight: +- 128 +- 384 +transformer.h.9.attn.c_proj.bias: +- 128 +transformer.h.9.attn.c_proj.weight: +- 128 +- 128 +transformer.h.9.ln_1.bias: +- 128 +transformer.h.9.ln_1.weight: +- 128 +transformer.h.9.ln_2.bias: +- 128 +transformer.h.9.ln_2.weight: +- 128 +transformer.h.9.mlp.c_fc.bias: +- 512 +transformer.h.9.mlp.c_fc.weight: +- 128 +- 512 +transformer.h.9.mlp.c_proj.bias: +- 128 +transformer.h.9.mlp.c_proj.weight: +- 512 +- 128 +transformer.ln_f.bias: +- 128 +transformer.ln_f.weight: +- 128 +transformer.wpe.weight: +- null +- 128 +transformer.wte.weight: +- null +- 128 diff --git a/mup-1b-100m-e3-full-gpt2lmmup.json b/mup-1b-100m-e3-full-gpt2lmmup.json deleted file mode 100644 index a78edc925629cc38586a2dd4fbe7d202aef93d99..0000000000000000000000000000000000000000 --- a/mup-1b-100m-e3-full-gpt2lmmup.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 10.503863334655762, "eval_runtime": 35656.762, "eval_samples_per_second": 4.751, "eval_steps_per_second": 0.148, "epoch": 0.01} \ No newline at end of file diff --git a/mup-1b-100m-e3-full.json b/mup-1b-100m-e3-full.json deleted file mode 100644 index a78edc925629cc38586a2dd4fbe7d202aef93d99..0000000000000000000000000000000000000000 --- a/mup-1b-100m-e3-full.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 10.503863334655762, "eval_runtime": 35656.762, "eval_samples_per_second": 4.751, "eval_steps_per_second": 0.148, "epoch": 0.01} \ No newline at end of file diff --git a/mup-1b-100m-e3/pytorch_model.bin b/mup-1b-100m-e3/pytorch_model.bin deleted file mode 100644 index 9ac2888d8366511a7708462277c11517b9fea70e..0000000000000000000000000000000000000000 --- a/mup-1b-100m-e3/pytorch_model.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:516a58a6402ecb11db1ae66007c6bd2dd06bca4d990a0ef70e9881c31f2e0525 -size 6080539805 diff --git a/mup-1b-100m-e3/training_args.bin b/mup-1b-100m-e3/training_args.bin deleted file mode 100644 index 54a08e088610bb59e41ede9a36eaa7a9f4294be3..0000000000000000000000000000000000000000 --- a/mup-1b-100m-e3/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:010e79df4f1f8fc111c9685a914acb53060ce3dcaf804d961ae7df3e2ee72c14 -size 3387 diff --git a/mup-1b-100m-e3/config.json b/mup-1b-100m/config.json similarity index 97% rename from mup-1b-100m-e3/config.json rename to mup-1b-100m/config.json index 0e0792cc38c0c0c0cb604686aa26e0fdf3b31296..22e5fa9afef6eef800fd3fbd083d331b4ce67a7c 100644 --- a/mup-1b-100m-e3/config.json +++ b/mup-1b-100m/config.json @@ -29,6 +29,6 @@ "summary_use_proj": true, "torch_dtype": "float32", "transformers_version": "4.25.1", - "use_cache": true, + "use_cache": false, "vocab_size": 50257 } diff --git a/mup-1b-100m/pytorch_model.bin b/mup-1b-100m/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..42b8db651f0478685ccf1a2af1b810ce5dc719a5 --- /dev/null +++ b/mup-1b-100m/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db917f7fa18c7c062716f3b195c484ed752da15765806f35f76b3e7f24d4c40a +size 6080539805 diff --git a/mup-1b-100m/training_args.bin b/mup-1b-100m/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..3b7eaae0df8fb5790b63c01d489cc0280a8820c4 --- /dev/null +++ b/mup-1b-100m/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1282be713f72d5c7a8476389a830fe32aa9d2ce295fc83219bc57e7b5ecf2e01 +size 3387 diff --git a/mup-200m-100m-e3-full-gpt2lmmup.json b/mup-200m-100m-e3-full-gpt2lmmup.json deleted file mode 100644 index f797a76e1d8bba8c1e07d8cbdb6b3a1e6f12129d..0000000000000000000000000000000000000000 --- a/mup-200m-100m-e3-full-gpt2lmmup.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 10.863893508911133, "eval_runtime": 35678.3953, "eval_samples_per_second": 4.748, "eval_steps_per_second": 0.148} \ No newline at end of file diff --git a/mup-200m-100m-e3-full.json b/mup-200m-100m-e3-full.json deleted file mode 100644 index f797a76e1d8bba8c1e07d8cbdb6b3a1e6f12129d..0000000000000000000000000000000000000000 --- a/mup-200m-100m-e3-full.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 10.863893508911133, "eval_runtime": 35678.3953, "eval_samples_per_second": 4.748, "eval_steps_per_second": 0.148} \ No newline at end of file diff --git a/mup-200m-100m-e3/config.json b/mup-200m-100m-e3/config.json deleted file mode 100644 index 3c89de94cdc339ba7df6839701d4fa04b955a065..0000000000000000000000000000000000000000 --- a/mup-200m-100m-e3/config.json +++ /dev/null @@ -1,34 +0,0 @@ -{ - "activation_function": "gelu_new", - "architectures": [ - "GPT2LMHeadModel" - ], - "attn_mult": 8.0, - "attn_pdrop": 0.1, - "bos_token_id": 50256, - "embd_pdrop": 0.1, - "eos_token_id": 50256, - "initializer_range": 0.1, - "intermediate_size": 12288, - "layer_norm_epsilon": 1e-05, - "model_type": "gpt2", - "n_embd": 3072, - "n_head": 64, - "n_inner": null, - "n_layer": 12, - "n_positions": 1024, - "num_layers": 12, - "reorder_and_upcast_attn": false, - "resid_pdrop": 0.1, - "scale_attn_by_inverse_layer_idx": false, - "scale_attn_weights": true, - "summary_activation": null, - "summary_first_dropout": 0.1, - "summary_proj_to_labels": true, - "summary_type": "cls_index", - "summary_use_proj": true, - "torch_dtype": "float32", - "transformers_version": "4.25.1", - "use_cache": true, - "vocab_size": 50257 -} diff --git a/mup-200m-100m-e3/pytorch_model.bin b/mup-200m-100m-e3/pytorch_model.bin deleted file mode 100644 index 02364d485b7fc5f2cfe85a21dafb6f526be07c2a..0000000000000000000000000000000000000000 --- a/mup-200m-100m-e3/pytorch_model.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:600250335c93538740793cc7e8a4682defb6f7e202a3641016cf4ee1754bbe75 -size 6080539805 diff --git a/mup-200m-100m-e3/training_args.bin b/mup-200m-100m-e3/training_args.bin deleted file mode 100644 index 0e8deb6debfd686b602f8dfe5d2c7b05a1f165bd..0000000000000000000000000000000000000000 --- a/mup-200m-100m-e3/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:8a9b9d007760086d2e55ee15759f906d50d344e948eb0bca87f0e434c9a4dd63 -size 3387 diff --git a/mup-3b5-100m-e3/config.json b/mup-200m-100m/checkpoint-100/config.json similarity index 92% rename from mup-3b5-100m-e3/config.json rename to mup-200m-100m/checkpoint-100/config.json index 7c89c4471675188d1ead6a2e0ef13de6e1ffa24d..c106051dfdf88b547dcd370556501888116ac330 100644 --- a/mup-3b5-100m-e3/config.json +++ b/mup-200m-100m/checkpoint-100/config.json @@ -9,11 +9,11 @@ "embd_pdrop": 0.1, "eos_token_id": 50256, "initializer_range": 0.01, - "intermediate_size": 18432, + "intermediate_size": 4096, "layer_norm_epsilon": 1e-05, "model_type": "gpt2", - "n_embd": 4608, - "n_head": 128, + "n_embd": 1024, + "n_head": 32, "n_inner": null, "n_layer": 12, "n_positions": 1024, diff --git a/mup-200m-100m/checkpoint-100/optimizer.pt b/mup-200m-100m/checkpoint-100/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..2ddc928f776d54d298a3a547c70a4e324d78808d --- /dev/null +++ b/mup-200m-100m/checkpoint-100/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:594bf76f545f0586abbd020a129730f10f1e43891d81f7de5e3728d84d1f283f +size 1629471237 diff --git a/mup-200m-100m/checkpoint-100/pytorch_model.bin b/mup-200m-100m/checkpoint-100/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..508b538d9e5104fc4f3740276434a783a697b404 --- /dev/null +++ b/mup-200m-100m/checkpoint-100/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3184aafc0c2d875a0eb5c978a5ed9ec2339f706fe9bd9a11bcc9bcd46ef6af66 +size 827312701 diff --git a/mup-200m-100m/checkpoint-100/rng_state.pth b/mup-200m-100m/checkpoint-100/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..ae9cceb6d5e49d67070bb8eff38f56cba6ff07b0 --- /dev/null +++ b/mup-200m-100m/checkpoint-100/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66ebba15e85e086af7736fdcf05e9cd682b1a2e05cbcf39415aaa86a3b656a0f +size 14575 diff --git a/mup-2b5-100m-e3/checkpoint-100/scaler.pt b/mup-200m-100m/checkpoint-100/scaler.pt similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/scaler.pt rename to mup-200m-100m/checkpoint-100/scaler.pt diff --git a/mup-200m-100m/checkpoint-100/scheduler.pt b/mup-200m-100m/checkpoint-100/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..97f13de8c2372bfdb5c176c2aac2773139dcbe45 --- /dev/null +++ b/mup-200m-100m/checkpoint-100/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac45f3b902e0fb39fb1b613f5d2434b6265aef8e4b3589db2a24c5a1ee39c2d4 +size 691 diff --git a/mup-2b-100m-e3/checkpoint-100/trainer_state.json b/mup-200m-100m/checkpoint-100/trainer_state.json similarity index 80% rename from mup-2b-100m-e3/checkpoint-100/trainer_state.json rename to mup-200m-100m/checkpoint-100/trainer_state.json index 6a2b08bfcbd1dd625bfcdbcacb742dad36486b1f..9c22b0b9f18f4e717c47e63e67c4d76fd52a03a9 100644 --- a/mup-2b-100m-e3/checkpoint-100/trainer_state.json +++ b/mup-200m-100m/checkpoint-100/trainer_state.json @@ -1,7 +1,7 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.2631010986526346, + "epoch": 0.2630929869275672, "global_step": 100, "is_hyper_param_search": false, "is_local_process_zero": true, @@ -9,7 +9,7 @@ "log_history": [], "max_steps": 380, "num_train_epochs": 1, - "total_flos": 2.50936282841088e+17, + "total_flos": 2.37748988411904e+16, "trial_name": null, "trial_params": null } diff --git a/mup-200m-100m/checkpoint-100/training_args.bin b/mup-200m-100m/checkpoint-100/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..85f2692e042d09f82f7a6adb1e7ab687b6baa7b6 --- /dev/null +++ b/mup-200m-100m/checkpoint-100/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eed287ec22ab538923d37b36905111de89f29c3d9044950da371af0f29af7ad2 +size 3387 diff --git a/mup-2b-100m-e3/checkpoint-200/config.json b/mup-200m-100m/checkpoint-200/config.json similarity index 89% rename from mup-2b-100m-e3/checkpoint-200/config.json rename to mup-200m-100m/checkpoint-200/config.json index 206494207d182569dd9eedb97a900ced82bdf150..c106051dfdf88b547dcd370556501888116ac330 100644 --- a/mup-2b-100m-e3/checkpoint-200/config.json +++ b/mup-200m-100m/checkpoint-200/config.json @@ -9,11 +9,11 @@ "embd_pdrop": 0.1, "eos_token_id": 50256, "initializer_range": 0.01, - "intermediate_size": 13312, + "intermediate_size": 4096, "layer_norm_epsilon": 1e-05, "model_type": "gpt2", - "n_embd": 3328, - "n_head": 128, + "n_embd": 1024, + "n_head": 32, "n_inner": null, "n_layer": 12, "n_positions": 1024, @@ -29,6 +29,6 @@ "summary_use_proj": true, "torch_dtype": "float32", "transformers_version": "4.25.1", - "use_cache": true, + "use_cache": false, "vocab_size": 50257 } diff --git a/mup-200m-100m/checkpoint-200/optimizer.pt b/mup-200m-100m/checkpoint-200/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..f25a8c7029036248145beb45a585d523f55fd6d7 --- /dev/null +++ b/mup-200m-100m/checkpoint-200/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22d39fe236529d8461887b25044dc49fb78fcf91a9c8d246a7e4b35ee6dc880d +size 1629471237 diff --git a/mup-200m-100m/checkpoint-200/pytorch_model.bin b/mup-200m-100m/checkpoint-200/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..75105ce2f413a1113f70105545b2f135b347ccc5 --- /dev/null +++ b/mup-200m-100m/checkpoint-200/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a932c129b0ecd3b2869b4cb68a7fd8c0325d26175279038e1778b6558ed5c40d +size 827312701 diff --git a/mup-200m-100m/checkpoint-200/rng_state.pth b/mup-200m-100m/checkpoint-200/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..25163a3f615e123afe6acbead1893960ac68f2c9 --- /dev/null +++ b/mup-200m-100m/checkpoint-200/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4851f3b302b7c4fee9d9420148364c058cf57804100694ae01ab586d33879732 +size 14575 diff --git a/mup-200m-100m/checkpoint-200/scaler.pt b/mup-200m-100m/checkpoint-200/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..87710d0ddc627c070366fcb3112b07dc60d97295 --- /dev/null +++ b/mup-200m-100m/checkpoint-200/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4fccf0f9be1bb8f24861e4393745b3e09cc2687125a69e3757955fb0f0925ea5 +size 557 diff --git a/mup-200m-100m/checkpoint-200/scheduler.pt b/mup-200m-100m/checkpoint-200/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..f309a574cc7ca8228ab7d18bc279ac7d470db205 --- /dev/null +++ b/mup-200m-100m/checkpoint-200/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88c8941f76ca00d1a3b4387d2503fd21325c149caf70f004ea30b3d20e440522 +size 691 diff --git a/mup-2b-100m-e3/checkpoint-200/trainer_state.json b/mup-200m-100m/checkpoint-200/trainer_state.json similarity index 80% rename from mup-2b-100m-e3/checkpoint-200/trainer_state.json rename to mup-200m-100m/checkpoint-200/trainer_state.json index 402a8b19a70c3f0a63f4f211c680c3f5bc507c7c..9e20aebfbd8a9a01122eba77d0dafd14c6188a4c 100644 --- a/mup-2b-100m-e3/checkpoint-200/trainer_state.json +++ b/mup-200m-100m/checkpoint-200/trainer_state.json @@ -1,7 +1,7 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.5262021973052692, + "epoch": 0.5261859738551344, "global_step": 200, "is_hyper_param_search": false, "is_local_process_zero": true, @@ -9,7 +9,7 @@ "log_history": [], "max_steps": 380, "num_train_epochs": 1, - "total_flos": 5.01872565682176e+17, + "total_flos": 4.75497976823808e+16, "trial_name": null, "trial_params": null } diff --git a/mup-200m-100m/checkpoint-200/training_args.bin b/mup-200m-100m/checkpoint-200/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..85f2692e042d09f82f7a6adb1e7ab687b6baa7b6 --- /dev/null +++ b/mup-200m-100m/checkpoint-200/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eed287ec22ab538923d37b36905111de89f29c3d9044950da371af0f29af7ad2 +size 3387 diff --git a/mup-200m-100m/checkpoint-300/config.json b/mup-200m-100m/checkpoint-300/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c106051dfdf88b547dcd370556501888116ac330 --- /dev/null +++ b/mup-200m-100m/checkpoint-300/config.json @@ -0,0 +1,34 @@ +{ + "activation_function": "gelu_new", + "architectures": [ + "GPT2LMHeadModel" + ], + "attn_mult": 8.0, + "attn_pdrop": 0.1, + "bos_token_id": 50256, + "embd_pdrop": 0.1, + "eos_token_id": 50256, + "initializer_range": 0.01, + "intermediate_size": 4096, + "layer_norm_epsilon": 1e-05, + "model_type": "gpt2", + "n_embd": 1024, + "n_head": 32, + "n_inner": null, + "n_layer": 12, + "n_positions": 1024, + "num_layers": 12, + "reorder_and_upcast_attn": false, + "resid_pdrop": 0.1, + "scale_attn_by_inverse_layer_idx": false, + "scale_attn_weights": true, + "summary_activation": null, + "summary_first_dropout": 0.1, + "summary_proj_to_labels": true, + "summary_type": "cls_index", + "summary_use_proj": true, + "torch_dtype": "float32", + "transformers_version": "4.25.1", + "use_cache": false, + "vocab_size": 50257 +} diff --git a/mup-200m-100m/checkpoint-300/optimizer.pt b/mup-200m-100m/checkpoint-300/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..71df9bfed0444a6ab6feaa3aed3ff617ac57d7fb --- /dev/null +++ b/mup-200m-100m/checkpoint-300/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:354296c51c24eca0644e80a3ac24b80b59e558aa504dffd9a191e8d3d0e03053 +size 1629471237 diff --git a/mup-200m-100m/checkpoint-300/pytorch_model.bin b/mup-200m-100m/checkpoint-300/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..5165814e2fde8b40016c1bd388d50768edfdd2e7 --- /dev/null +++ b/mup-200m-100m/checkpoint-300/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87a43819750fecf47f4789e43fc5196dc2175cffef8785d0d916b9f439adad0a +size 827312701 diff --git a/mup-200m-100m/checkpoint-300/rng_state.pth b/mup-200m-100m/checkpoint-300/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..faaebb5af425ee88845db69079f330c73f50b52a --- /dev/null +++ b/mup-200m-100m/checkpoint-300/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08950dc609e5df5b063c350f814f906ffba4fb0870677e17789bafc766cec4b7 +size 14575 diff --git a/mup-200m-100m/checkpoint-300/scaler.pt b/mup-200m-100m/checkpoint-300/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..96d1b8ea2cbad321ffe9c10840a99bb1bcef18f5 --- /dev/null +++ b/mup-200m-100m/checkpoint-300/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efafd90182e3d39d1b7c4a686f86e5913f5abc094dc3e2f827a6d479c6cef247 +size 557 diff --git a/mup-200m-100m/checkpoint-300/scheduler.pt b/mup-200m-100m/checkpoint-300/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..9e0c7b50c1389d2403dd9e572397df6e92f54f51 --- /dev/null +++ b/mup-200m-100m/checkpoint-300/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f22762c604e0f6d64b39e692dc2d5b773eb950be2aea906ecde2da50f125b1ad +size 691 diff --git a/mup-3b-100m-e3/checkpoint-100/trainer_state.json b/mup-200m-100m/checkpoint-300/trainer_state.json similarity index 69% rename from mup-3b-100m-e3/checkpoint-100/trainer_state.json rename to mup-200m-100m/checkpoint-300/trainer_state.json index 2e949b9ca48b5958c0ab8a9bdb68df2951723738..cf06e326ddb04069ed7461ad76a7a302a01f5cbb 100644 --- a/mup-3b-100m-e3/checkpoint-100/trainer_state.json +++ b/mup-200m-100m/checkpoint-300/trainer_state.json @@ -1,15 +1,15 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.5262021973052692, - "global_step": 100, + "epoch": 0.7892789607827017, + "global_step": 300, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [], - "max_steps": 190, + "max_steps": 380, "num_train_epochs": 1, - "total_flos": 8.581652284440576e+17, + "total_flos": 7.13246965235712e+16, "trial_name": null, "trial_params": null } diff --git a/mup-200m-100m/checkpoint-300/training_args.bin b/mup-200m-100m/checkpoint-300/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..85f2692e042d09f82f7a6adb1e7ab687b6baa7b6 --- /dev/null +++ b/mup-200m-100m/checkpoint-300/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eed287ec22ab538923d37b36905111de89f29c3d9044950da371af0f29af7ad2 +size 3387 diff --git a/mup-200m-100m/config.json b/mup-200m-100m/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c106051dfdf88b547dcd370556501888116ac330 --- /dev/null +++ b/mup-200m-100m/config.json @@ -0,0 +1,34 @@ +{ + "activation_function": "gelu_new", + "architectures": [ + "GPT2LMHeadModel" + ], + "attn_mult": 8.0, + "attn_pdrop": 0.1, + "bos_token_id": 50256, + "embd_pdrop": 0.1, + "eos_token_id": 50256, + "initializer_range": 0.01, + "intermediate_size": 4096, + "layer_norm_epsilon": 1e-05, + "model_type": "gpt2", + "n_embd": 1024, + "n_head": 32, + "n_inner": null, + "n_layer": 12, + "n_positions": 1024, + "num_layers": 12, + "reorder_and_upcast_attn": false, + "resid_pdrop": 0.1, + "scale_attn_by_inverse_layer_idx": false, + "scale_attn_weights": true, + "summary_activation": null, + "summary_first_dropout": 0.1, + "summary_proj_to_labels": true, + "summary_type": "cls_index", + "summary_use_proj": true, + "torch_dtype": "float32", + "transformers_version": "4.25.1", + "use_cache": false, + "vocab_size": 50257 +} diff --git a/mup-200m-100m/pytorch_model.bin b/mup-200m-100m/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..6e5863ff503271c9188af9ff2b185ab51946049f --- /dev/null +++ b/mup-200m-100m/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fce93a7c72644099928e5c71d3c690ab11190cce10bb69a429817748890e896d +size 827312701 diff --git a/mup-200m-100m/training_args.bin b/mup-200m-100m/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..85f2692e042d09f82f7a6adb1e7ab687b6baa7b6 --- /dev/null +++ b/mup-200m-100m/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eed287ec22ab538923d37b36905111de89f29c3d9044950da371af0f29af7ad2 +size 3387 diff --git a/mup-2b-100m-e3-full-gpt2lmmup.json b/mup-2b-100m-e3-full-gpt2lmmup.json deleted file mode 100644 index 21cd2466f8af116d4bb9a8164c00e68695d24e05..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3-full-gpt2lmmup.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 9.270681381225586, "eval_runtime": 46206.7705, "eval_samples_per_second": 3.666, "eval_steps_per_second": 0.229} \ No newline at end of file diff --git a/mup-2b-100m-e3-full.json b/mup-2b-100m-e3-full.json deleted file mode 100644 index 21cd2466f8af116d4bb9a8164c00e68695d24e05..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3-full.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 9.270681381225586, "eval_runtime": 46206.7705, "eval_samples_per_second": 3.666, "eval_steps_per_second": 0.229} \ No newline at end of file diff --git a/mup-2b-100m-e3/checkpoint-100/optimizer.pt b/mup-2b-100m-e3/checkpoint-100/optimizer.pt deleted file mode 100644 index 91c515eca1ef3b95304cef2498c5529b7b08ce31..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-100/optimizer.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:859a1cea6ab663b2651f6b47d6e85d9acb2a67b1ddf7200f7e5c350ab22c1f3b -size 14128712873 diff --git a/mup-2b-100m-e3/checkpoint-100/pytorch_model.bin b/mup-2b-100m-e3/checkpoint-100/pytorch_model.bin deleted file mode 100644 index f3f8eea86f0476580681d90ce09683d450abdcfb..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-100/pytorch_model.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:9aff050f6759cb9c9bd9fb82d5908dc746bd5fa9f0754b969286830370959df2 -size 7076932005 diff --git a/mup-2b-100m-e3/checkpoint-100/rng_state.pth b/mup-2b-100m-e3/checkpoint-100/rng_state.pth deleted file mode 100644 index 213cb727666a3e0291a137e9f014e650f3be8706..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-100/rng_state.pth +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:4c85e1a7fcee9f0a4e6f8ebf1dc4a63a8c96bd1dd0a67ff0ff15c506aae95304 -size 14575 diff --git a/mup-2b-100m-e3/checkpoint-100/scheduler.pt b/mup-2b-100m-e3/checkpoint-100/scheduler.pt deleted file mode 100644 index 4e56abbb618df6be14e8e38fac0265d36a57a011..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-100/scheduler.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f41127ce2e7b57031bf845923aadf69d1d932623d3cc306538c8cffe3f95b12f -size 691 diff --git a/mup-2b-100m-e3/checkpoint-100/training_args.bin b/mup-2b-100m-e3/checkpoint-100/training_args.bin deleted file mode 100644 index aa4c4b243cd83ca72ae78bb76bbca13363e894dc..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-100/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a529bfebc81b3f76599a2706a7dedaa2e159f90434d17d7113ba289ad90ba128 -size 3387 diff --git a/mup-2b-100m-e3/checkpoint-200/optimizer.pt b/mup-2b-100m-e3/checkpoint-200/optimizer.pt deleted file mode 100644 index caba7a46c14ad8b4691520849712958256926953..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-200/optimizer.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:fc8643fa7f495fb7235042b7a64c424947f35a3834c2e7ecaf81a5faff11ddb3 -size 14128712873 diff --git a/mup-2b-100m-e3/checkpoint-200/pytorch_model.bin b/mup-2b-100m-e3/checkpoint-200/pytorch_model.bin deleted file mode 100644 index 238d32bb7eca9f593a3bac0b1c84b9c3e6aa3bb4..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-200/pytorch_model.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:64362990e2eaa64481e967769bacc656341aa1ab8490f6c9a62abaa928ac7f9d -size 7076932005 diff --git a/mup-2b-100m-e3/checkpoint-200/rng_state.pth b/mup-2b-100m-e3/checkpoint-200/rng_state.pth deleted file mode 100644 index 9c6dbf95eb9baebca965add5859888c56c8ed597..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-200/rng_state.pth +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:850a75d03b68709ebdb3b0429ccf1c2bfbc264ba389a899cc590928049b771ed -size 14575 diff --git a/mup-2b-100m-e3/checkpoint-200/scheduler.pt b/mup-2b-100m-e3/checkpoint-200/scheduler.pt deleted file mode 100644 index f51d475b1ca12195a990369a485e37ac3cfffe72..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-200/scheduler.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:7ff11a668fd015bf8fec7ef706376a383794600ae7f4f86f3b9aeffe27fd39be -size 691 diff --git a/mup-2b-100m-e3/checkpoint-200/training_args.bin b/mup-2b-100m-e3/checkpoint-200/training_args.bin deleted file mode 100644 index aa4c4b243cd83ca72ae78bb76bbca13363e894dc..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-200/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a529bfebc81b3f76599a2706a7dedaa2e159f90434d17d7113ba289ad90ba128 -size 3387 diff --git a/mup-2b-100m-e3/checkpoint-300/optimizer.pt b/mup-2b-100m-e3/checkpoint-300/optimizer.pt deleted file mode 100644 index 7a510a1b49a2b71d0a9e735c7e5c51e4e337d82c..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-300/optimizer.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:83e214ea9a2859b68902931336a551a4a675fe79c227fa5b6bf330b7d270514b -size 14128712873 diff --git a/mup-2b-100m-e3/checkpoint-300/pytorch_model.bin b/mup-2b-100m-e3/checkpoint-300/pytorch_model.bin deleted file mode 100644 index d9af1b97b642155ed4b934411e3d2d34bcba0f5f..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-300/pytorch_model.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:b0b6c22fcc540890eee68ac4649802608865e27c9584d1917a079d4034b8ab41 -size 7076932005 diff --git a/mup-2b-100m-e3/checkpoint-300/scheduler.pt b/mup-2b-100m-e3/checkpoint-300/scheduler.pt deleted file mode 100644 index 9f8a9962b2b1866f629b67355349edba2c7229fb..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-300/scheduler.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:30fd152d86dafeffe2455270e22218f0b6cab42a3c9b31a01b2bc6d8ceadc7c4 -size 691 diff --git a/mup-2b-100m-e3/checkpoint-300/training_args.bin b/mup-2b-100m-e3/checkpoint-300/training_args.bin deleted file mode 100644 index aa4c4b243cd83ca72ae78bb76bbca13363e894dc..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/checkpoint-300/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a529bfebc81b3f76599a2706a7dedaa2e159f90434d17d7113ba289ad90ba128 -size 3387 diff --git a/mup-2b-100m-e3/pytorch_model.bin b/mup-2b-100m-e3/pytorch_model.bin deleted file mode 100644 index 67bf658114f0bbb790186b7f5b17e9c3b8033061..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/pytorch_model.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:9c72ffdc020dcc3f482979a1b900d14d4832f0f3f07845f1feaba439e3d93b68 -size 7076932005 diff --git a/mup-2b-100m-e3/training_args.bin b/mup-2b-100m-e3/training_args.bin deleted file mode 100644 index 81c35996da1f93cb7bd14f67fd12d176a6fcf80c..0000000000000000000000000000000000000000 --- a/mup-2b-100m-e3/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c0360c4e8f458bf1a55bfbc7c49d3cd1b51ecf6c15f5985ea0900cb0d4d27bfb -size 3387 diff --git a/mup-2b-100m-e3/checkpoint-300/config.json b/mup-2b-100m/config.json similarity index 97% rename from mup-2b-100m-e3/checkpoint-300/config.json rename to mup-2b-100m/config.json index 206494207d182569dd9eedb97a900ced82bdf150..35994f9fd0c03bbdb147496d08b3c74c4d4b7b6e 100644 --- a/mup-2b-100m-e3/checkpoint-300/config.json +++ b/mup-2b-100m/config.json @@ -29,6 +29,6 @@ "summary_use_proj": true, "torch_dtype": "float32", "transformers_version": "4.25.1", - "use_cache": true, + "use_cache": false, "vocab_size": 50257 } diff --git a/mup-2b-100m/pytorch_model.bin b/mup-2b-100m/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..5dbaf70d1f2218cb195c820128c2e1e9e0c9e7f4 --- /dev/null +++ b/mup-2b-100m/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78adb59edc6f9e1259d95cfe9c9c4ff789abf403abcb72cd40a294b078e51653 +size 7076932005 diff --git a/mup-2b-100m/training_args.bin b/mup-2b-100m/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..28d9426e56dc53b35785afca211448ba4847c5d8 --- /dev/null +++ b/mup-2b-100m/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97eee37826c3cd5dd0f1f8ec1f9fdde2d51145e25bafb34ffd1a872005f925d6 +size 3387 diff --git a/mup-2b5-100m-e3-full-gpt2lmmup.json b/mup-2b5-100m-e3-full-gpt2lmmup.json deleted file mode 100644 index babff691199e46285cb34ffe3f1dd30da120e317..0000000000000000000000000000000000000000 --- a/mup-2b5-100m-e3-full-gpt2lmmup.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 6.969681739807129, "eval_runtime": 14651.8437, "eval_samples_per_second": 11.563, "eval_steps_per_second": 0.723, "epoch": 1.0} \ No newline at end of file diff --git a/mup-2b5-100m-e3-full.json b/mup-2b5-100m-e3-full.json deleted file mode 100644 index babff691199e46285cb34ffe3f1dd30da120e317..0000000000000000000000000000000000000000 --- a/mup-2b5-100m-e3-full.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 6.969681739807129, "eval_runtime": 14651.8437, "eval_samples_per_second": 11.563, "eval_steps_per_second": 0.723, "epoch": 1.0} \ No newline at end of file diff --git a/mup-2b5-100m-e3/checkpoint-100/config.json b/mup-2b5-100m/checkpoint-100/config.json similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/config.json rename to mup-2b5-100m/checkpoint-100/config.json diff --git a/mup-2b5-100m-e3/checkpoint-100/optimizer.pt b/mup-2b5-100m/checkpoint-100/optimizer.pt similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/optimizer.pt rename to mup-2b5-100m/checkpoint-100/optimizer.pt diff --git a/mup-2b5-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin b/mup-2b5-100m/checkpoint-100/pytorch_model-00001-of-00002.bin similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin rename to mup-2b5-100m/checkpoint-100/pytorch_model-00001-of-00002.bin diff --git a/mup-2b5-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin b/mup-2b5-100m/checkpoint-100/pytorch_model-00002-of-00002.bin similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin rename to mup-2b5-100m/checkpoint-100/pytorch_model-00002-of-00002.bin diff --git a/mup-2b5-100m-e3/checkpoint-100/pytorch_model.bin.index.json b/mup-2b5-100m/checkpoint-100/pytorch_model.bin.index.json similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/pytorch_model.bin.index.json rename to mup-2b5-100m/checkpoint-100/pytorch_model.bin.index.json diff --git a/mup-2b5-100m-e3/checkpoint-100/rng_state.pth b/mup-2b5-100m/checkpoint-100/rng_state.pth similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/rng_state.pth rename to mup-2b5-100m/checkpoint-100/rng_state.pth diff --git a/mup-3b-100m-e3/checkpoint-100/scaler.pt b/mup-2b5-100m/checkpoint-100/scaler.pt similarity index 100% rename from mup-3b-100m-e3/checkpoint-100/scaler.pt rename to mup-2b5-100m/checkpoint-100/scaler.pt diff --git a/mup-2b5-100m-e3/checkpoint-100/scheduler.pt b/mup-2b5-100m/checkpoint-100/scheduler.pt similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/scheduler.pt rename to mup-2b5-100m/checkpoint-100/scheduler.pt diff --git a/mup-2b5-100m-e3/checkpoint-100/trainer_state.json b/mup-2b5-100m/checkpoint-100/trainer_state.json similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/trainer_state.json rename to mup-2b5-100m/checkpoint-100/trainer_state.json diff --git a/mup-2b5-100m-e3/checkpoint-100/training_args.bin b/mup-2b5-100m/checkpoint-100/training_args.bin similarity index 100% rename from mup-2b5-100m-e3/checkpoint-100/training_args.bin rename to mup-2b5-100m/checkpoint-100/training_args.bin diff --git a/mup-2b5-100m-e3/config.json b/mup-2b5-100m/config.json similarity index 100% rename from mup-2b5-100m-e3/config.json rename to mup-2b5-100m/config.json diff --git a/mup-2b5-100m-e3/pytorch_model-00001-of-00002.bin b/mup-2b5-100m/pytorch_model-00001-of-00002.bin similarity index 100% rename from mup-2b5-100m-e3/pytorch_model-00001-of-00002.bin rename to mup-2b5-100m/pytorch_model-00001-of-00002.bin diff --git a/mup-2b5-100m-e3/pytorch_model-00002-of-00002.bin b/mup-2b5-100m/pytorch_model-00002-of-00002.bin similarity index 100% rename from mup-2b5-100m-e3/pytorch_model-00002-of-00002.bin rename to mup-2b5-100m/pytorch_model-00002-of-00002.bin diff --git a/mup-2b5-100m-e3/pytorch_model.bin.index.json b/mup-2b5-100m/pytorch_model.bin.index.json similarity index 100% rename from mup-2b5-100m-e3/pytorch_model.bin.index.json rename to mup-2b5-100m/pytorch_model.bin.index.json diff --git a/mup-2b5-100m-e3/training_args.bin b/mup-2b5-100m/training_args.bin similarity index 100% rename from mup-2b5-100m-e3/training_args.bin rename to mup-2b5-100m/training_args.bin diff --git a/mup-3b-100m-e3-full-gpt2lmmup.json b/mup-3b-100m-e3-full-gpt2lmmup.json deleted file mode 100644 index 5e84b5ab3a1c1e1451bcf755e88b781825d921cc..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3-full-gpt2lmmup.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 6.870556831359863, "eval_runtime": 16069.3588, "eval_samples_per_second": 10.543, "eval_steps_per_second": 0.659, "epoch": 1.0} \ No newline at end of file diff --git a/mup-3b-100m-e3-full.json b/mup-3b-100m-e3-full.json deleted file mode 100644 index 5e84b5ab3a1c1e1451bcf755e88b781825d921cc..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3-full.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 6.870556831359863, "eval_runtime": 16069.3588, "eval_samples_per_second": 10.543, "eval_steps_per_second": 0.659, "epoch": 1.0} \ No newline at end of file diff --git a/mup-3b-100m-e3/checkpoint-100/optimizer.pt b/mup-3b-100m-e3/checkpoint-100/optimizer.pt deleted file mode 100644 index f1a86f76a2fc4d707f00d34ee2d52a0099f7c0af..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3/checkpoint-100/optimizer.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:590f451475b76a5bb9001c773d83e4b88c75c838fed4534df9f033eadba5470b -size 23609798041 diff --git a/mup-3b-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin b/mup-3b-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin deleted file mode 100644 index a50872e8be7d2335df1e709bb3ca3a44c0b589e6..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:61159c7f524a355b65044bed24541028017265e83427b9c19a9a1a75c006948f -size 9997736349 diff --git a/mup-3b-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin b/mup-3b-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin deleted file mode 100644 index 51c827e9e420f8fd589a38a9fa44f34b85b1ade0..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:639deb25bc2c13466f03e85aab820a2c107f03982379eaf3ce473dd5f4c38a9e -size 2694614901 diff --git a/mup-3b-100m-e3/checkpoint-100/rng_state.pth b/mup-3b-100m-e3/checkpoint-100/rng_state.pth deleted file mode 100644 index c365ed9430e8e47e1ee71469df63d8da2a549378..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3/checkpoint-100/rng_state.pth +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bccf3cda0d933f5f7bb7859163a2e83eefeef8218429908f35fc718e326740df -size 14575 diff --git a/mup-3b-100m-e3/checkpoint-100/scheduler.pt b/mup-3b-100m-e3/checkpoint-100/scheduler.pt deleted file mode 100644 index 98c9eb65a229debc77fd68453d69b24aa4d94a18..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3/checkpoint-100/scheduler.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:96cd9e71e6cb99c3e129d0a33d4fa0d5f079e1ff04df09a4e8e7d88ca6a01a02 -size 691 diff --git a/mup-3b-100m-e3/config.json b/mup-3b-100m-e3/config.json deleted file mode 100644 index cc3939892f1fd961b9edefe961a90d83cc981ada..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3/config.json +++ /dev/null @@ -1,34 +0,0 @@ -{ - "activation_function": "gelu_new", - "architectures": [ - "GPT2LMHeadModel" - ], - "attn_mult": 8.0, - "attn_pdrop": 0.1, - "bos_token_id": 50256, - "embd_pdrop": 0.1, - "eos_token_id": 50256, - "initializer_range": 0.01, - "intermediate_size": 17408, - "layer_norm_epsilon": 1e-05, - "model_type": "gpt2", - "n_embd": 4352, - "n_head": 128, - "n_inner": null, - "n_layer": 12, - "n_positions": 1024, - "num_layers": 12, - "reorder_and_upcast_attn": false, - "resid_pdrop": 0.1, - "scale_attn_by_inverse_layer_idx": false, - "scale_attn_weights": true, - "summary_activation": null, - "summary_first_dropout": 0.1, - "summary_proj_to_labels": true, - "summary_type": "cls_index", - "summary_use_proj": true, - "torch_dtype": "float32", - "transformers_version": "4.25.1", - "use_cache": true, - "vocab_size": 50257 -} diff --git a/mup-3b-100m-e3/training_args.bin b/mup-3b-100m-e3/training_args.bin deleted file mode 100644 index 5fb4f25ccac324f9937e08ecab989b76f2ff85d3..0000000000000000000000000000000000000000 --- a/mup-3b-100m-e3/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:ee5a7ca6b12b62d3ee6f578e53b23b4113c84732ffaa9d8d7cff6999532bf537 -size 3387 diff --git a/mup-3b-100m-e3/checkpoint-100/config.json b/mup-3b-100m/config.json similarity index 100% rename from mup-3b-100m-e3/checkpoint-100/config.json rename to mup-3b-100m/config.json diff --git a/mup-3b-100m-e3/pytorch_model-00001-of-00002.bin b/mup-3b-100m/pytorch_model-00001-of-00002.bin similarity index 100% rename from mup-3b-100m-e3/pytorch_model-00001-of-00002.bin rename to mup-3b-100m/pytorch_model-00001-of-00002.bin diff --git a/mup-3b-100m-e3/pytorch_model-00002-of-00002.bin b/mup-3b-100m/pytorch_model-00002-of-00002.bin similarity index 100% rename from mup-3b-100m-e3/pytorch_model-00002-of-00002.bin rename to mup-3b-100m/pytorch_model-00002-of-00002.bin diff --git a/mup-3b-100m-e3/checkpoint-100/pytorch_model.bin.index.json b/mup-3b-100m/pytorch_model.bin.index.json similarity index 100% rename from mup-3b-100m-e3/checkpoint-100/pytorch_model.bin.index.json rename to mup-3b-100m/pytorch_model.bin.index.json diff --git a/mup-3b-100m-e3/checkpoint-100/training_args.bin b/mup-3b-100m/training_args.bin similarity index 100% rename from mup-3b-100m-e3/checkpoint-100/training_args.bin rename to mup-3b-100m/training_args.bin diff --git a/mup-3b5-100m-e3-full-gpt2lmmup.json b/mup-3b5-100m-e3-full-gpt2lmmup.json deleted file mode 100644 index cc5c83e15d9de6ee14c40919e630303e108b7354..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3-full-gpt2lmmup.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 6.883823394775391, "eval_runtime": 16717.2968, "eval_samples_per_second": 10.134, "eval_steps_per_second": 0.633, "epoch": 1.0} \ No newline at end of file diff --git a/mup-3b5-100m-e3-full.json b/mup-3b5-100m-e3-full.json deleted file mode 100644 index cc5c83e15d9de6ee14c40919e630303e108b7354..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3-full.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 6.883823394775391, "eval_runtime": 16717.2968, "eval_samples_per_second": 10.134, "eval_steps_per_second": 0.633, "epoch": 1.0} \ No newline at end of file diff --git a/mup-3b5-100m-e3/checkpoint-100/optimizer.pt b/mup-3b5-100m-e3/checkpoint-100/optimizer.pt deleted file mode 100644 index 04f5e65994098df964934c67931a6237a49d8fb4..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3/checkpoint-100/optimizer.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a2a98488d55c99a2547d7fd062162a3229860ed14ad2044c855554eb3344ebc4 -size 26357556681 diff --git a/mup-3b5-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin b/mup-3b5-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin deleted file mode 100644 index 401a9631f60e217e4b01a488184e8d2afc7acc7e..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3/checkpoint-100/pytorch_model-00001-of-00002.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:60fbb7266991536dcb61ef1a8f8e2841b176b5a086bbd028f9f47beb55486c59 -size 9790034561 diff --git a/mup-3b5-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin b/mup-3b5-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin deleted file mode 100644 index 28cf95cf2c7e0d1b605c3a06326766bfc9e47031..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3/checkpoint-100/pytorch_model-00002-of-00002.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:28ce0c918e97156f9d021dea864ab763be1761c08273bf73310804cca41f7c3e -size 4327659005 diff --git a/mup-3b5-100m-e3/checkpoint-100/rng_state.pth b/mup-3b5-100m-e3/checkpoint-100/rng_state.pth deleted file mode 100644 index c365ed9430e8e47e1ee71469df63d8da2a549378..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3/checkpoint-100/rng_state.pth +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bccf3cda0d933f5f7bb7859163a2e83eefeef8218429908f35fc718e326740df -size 14575 diff --git a/mup-3b5-100m-e3/checkpoint-100/scheduler.pt b/mup-3b5-100m-e3/checkpoint-100/scheduler.pt deleted file mode 100644 index 325b9781497a3c777bc2ada8a6eb4623bc7dfb02..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3/checkpoint-100/scheduler.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:08e7910132718b7b09982cd5a6bc29605e45eafd7b09d4d1f772b55fa781189b -size 691 diff --git a/mup-3b5-100m-e3/pytorch_model.bin.index.json b/mup-3b5-100m-e3/pytorch_model.bin.index.json deleted file mode 100644 index a453a1bab2e072a7aa46eaaaba4288a5116263c9..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3/pytorch_model.bin.index.json +++ /dev/null @@ -1,180 +0,0 @@ -{ - "metadata": { - "total_size": 14117634096 - }, - "weight_map": { - "lm_head.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.0.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.0.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.10.attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.c_attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.c_attn.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.masked_bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.ln_1.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.ln_1.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.ln_2.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.ln_2.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_attn.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.masked_bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_1.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_1.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_2.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_2.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.mlp.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.2.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.2.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.3.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.4.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.5.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.6.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.7.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.attn.masked_bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.ln_1.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.ln_2.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.ln_2.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.8.mlp.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.8.mlp.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.attn.c_attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.attn.c_attn.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.attn.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.attn.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.attn.masked_bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.ln_1.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.ln_1.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.ln_2.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.ln_2.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.mlp.c_fc.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.mlp.c_fc.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.mlp.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.9.mlp.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.ln_f.bias": "pytorch_model-00002-of-00002.bin", - "transformer.ln_f.weight": "pytorch_model-00002-of-00002.bin", - "transformer.wpe.weight": "pytorch_model-00001-of-00002.bin", - "transformer.wte.weight": "pytorch_model-00001-of-00002.bin" - } -} diff --git a/mup-3b5-100m-e3/training_args.bin b/mup-3b5-100m-e3/training_args.bin deleted file mode 100644 index 42a506cb5d02bc0d96e7a45b0f22d47e16001352..0000000000000000000000000000000000000000 --- a/mup-3b5-100m-e3/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bfa71a54b9260bfd5ca844c5e130f22f2225f6f0e829b91bc766959a5d8df96d -size 3387 diff --git a/mup-3b5-100m-e3/checkpoint-100/config.json b/mup-3b5-100m/config.json similarity index 100% rename from mup-3b5-100m-e3/checkpoint-100/config.json rename to mup-3b5-100m/config.json diff --git a/mup-3b5-100m-e3/pytorch_model-00001-of-00002.bin b/mup-3b5-100m/pytorch_model-00001-of-00002.bin similarity index 100% rename from mup-3b5-100m-e3/pytorch_model-00001-of-00002.bin rename to mup-3b5-100m/pytorch_model-00001-of-00002.bin diff --git a/mup-3b5-100m-e3/pytorch_model-00002-of-00002.bin b/mup-3b5-100m/pytorch_model-00002-of-00002.bin similarity index 100% rename from mup-3b5-100m-e3/pytorch_model-00002-of-00002.bin rename to mup-3b5-100m/pytorch_model-00002-of-00002.bin diff --git a/mup-3b5-100m-e3/checkpoint-100/pytorch_model.bin.index.json b/mup-3b5-100m/pytorch_model.bin.index.json similarity index 100% rename from mup-3b5-100m-e3/checkpoint-100/pytorch_model.bin.index.json rename to mup-3b5-100m/pytorch_model.bin.index.json diff --git a/mup-3b5-100m-e3/checkpoint-100/training_args.bin b/mup-3b5-100m/training_args.bin similarity index 100% rename from mup-3b5-100m-e3/checkpoint-100/training_args.bin rename to mup-3b5-100m/training_args.bin diff --git a/mup-800m-100m-e3-full-gpt2lmmup.json b/mup-800m-100m-e3-full-gpt2lmmup.json deleted file mode 100644 index c84e580b0beb62c82ab3f2ad13f12310969b5bd8..0000000000000000000000000000000000000000 --- a/mup-800m-100m-e3-full-gpt2lmmup.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 10.501672744750977, "eval_runtime": 19631.6072, "eval_samples_per_second": 8.63, "eval_steps_per_second": 0.27, "epoch": 0.01} \ No newline at end of file diff --git a/mup-800m-100m-e3-full.json b/mup-800m-100m-e3-full.json deleted file mode 100644 index c84e580b0beb62c82ab3f2ad13f12310969b5bd8..0000000000000000000000000000000000000000 --- a/mup-800m-100m-e3-full.json +++ /dev/null @@ -1 +0,0 @@ -{"eval_loss": 10.501672744750977, "eval_runtime": 19631.6072, "eval_samples_per_second": 8.63, "eval_steps_per_second": 0.27, "epoch": 0.01} \ No newline at end of file diff --git a/mup-800m-100m-e3/pytorch_model.bin b/mup-800m-100m-e3/pytorch_model.bin deleted file mode 100644 index 59bbb9b2af55b58d7ee7a8d4758b4c8fac164a06..0000000000000000000000000000000000000000 --- a/mup-800m-100m-e3/pytorch_model.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bc2ac8961f093b4652a4b03576267273027eefef683cfb5d1ef54a6e2846a48b -size 2849946173 diff --git a/mup-800m-100m-e3/training_args.bin b/mup-800m-100m-e3/training_args.bin deleted file mode 100644 index e6d2375da6c37647db0e48a9ca43f87c50ce8f57..0000000000000000000000000000000000000000 --- a/mup-800m-100m-e3/training_args.bin +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:b4bd735f045412800863daa4ff54012c625b362504d08515303228d130e20641 -size 3387 diff --git a/mup-800m-100m-e3/config.json b/mup-800m-100m/checkpoint-100/config.json similarity index 97% rename from mup-800m-100m-e3/config.json rename to mup-800m-100m/checkpoint-100/config.json index d47a25b9078f5cf21d1fc261182c895f660a490c..cd4fa9efea20031e766b401a8f2b6481d482fde0 100644 --- a/mup-800m-100m-e3/config.json +++ b/mup-800m-100m/checkpoint-100/config.json @@ -29,6 +29,6 @@ "summary_use_proj": true, "torch_dtype": "float32", "transformers_version": "4.25.1", - "use_cache": true, + "use_cache": false, "vocab_size": 50257 } diff --git a/mup-800m-100m/checkpoint-100/optimizer.pt b/mup-800m-100m/checkpoint-100/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..be11187d7724cf8b840c52a9b6ce076750870c69 --- /dev/null +++ b/mup-800m-100m/checkpoint-100/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eca00386bcf18be437e9ec6633fb6ffa939e1717839d53a00ce89d8defcb902d +size 5674738681 diff --git a/mup-800m-100m/checkpoint-100/pytorch_model.bin b/mup-800m-100m/checkpoint-100/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..8ea0446cae849114c1333bb74bda6f7de7baa55e --- /dev/null +++ b/mup-800m-100m/checkpoint-100/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0d754dbe9510e954190b5fd2803731a9635016a0d4b486fad5246ee9f143e19 +size 2849946173 diff --git a/mup-800m-100m/checkpoint-100/rng_state.pth b/mup-800m-100m/checkpoint-100/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..25163a3f615e123afe6acbead1893960ac68f2c9 --- /dev/null +++ b/mup-800m-100m/checkpoint-100/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4851f3b302b7c4fee9d9420148364c058cf57804100694ae01ab586d33879732 +size 14575 diff --git a/mup-3b5-100m-e3/checkpoint-100/scaler.pt b/mup-800m-100m/checkpoint-100/scaler.pt similarity index 100% rename from mup-3b5-100m-e3/checkpoint-100/scaler.pt rename to mup-800m-100m/checkpoint-100/scaler.pt diff --git a/mup-800m-100m/checkpoint-100/scheduler.pt b/mup-800m-100m/checkpoint-100/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..6b9e1cb33273684e329092efcd1e334be1afbeb9 --- /dev/null +++ b/mup-800m-100m/checkpoint-100/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:edcfb7c3e5387cd27d04acc4237fb579df96222c3fdd9d48479119909b5977e5 +size 691 diff --git a/mup-3b5-100m-e3/checkpoint-100/trainer_state.json b/mup-800m-100m/checkpoint-100/trainer_state.json similarity index 75% rename from mup-3b5-100m-e3/checkpoint-100/trainer_state.json rename to mup-800m-100m/checkpoint-100/trainer_state.json index 0f7c7676c40846a58d8c85b9591799edde0c942f..dc938ec2337ab173bb50e591be5d2d92182cbd10 100644 --- a/mup-3b5-100m-e3/checkpoint-100/trainer_state.json +++ b/mup-800m-100m/checkpoint-100/trainer_state.json @@ -1,15 +1,15 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.5262021973052692, + "epoch": 0.26309839468870116, "global_step": 100, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [], - "max_steps": 190, + "max_steps": 380, "num_train_epochs": 1, - "total_flos": 9.62081801109504e+17, + "total_flos": 9.5048700002304e+16, "trial_name": null, "trial_params": null } diff --git a/mup-800m-100m/checkpoint-100/training_args.bin b/mup-800m-100m/checkpoint-100/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..507802263d44f3aa55e5dba70d42b153113db0d9 --- /dev/null +++ b/mup-800m-100m/checkpoint-100/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e10f7310880f2f2fc7fc715f393057fe3a9edc2aef2788ec38f44a4faa96083 +size 3387 diff --git a/mup-800m-100m/checkpoint-200/config.json b/mup-800m-100m/checkpoint-200/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cd4fa9efea20031e766b401a8f2b6481d482fde0 --- /dev/null +++ b/mup-800m-100m/checkpoint-200/config.json @@ -0,0 +1,34 @@ +{ + "activation_function": "gelu_new", + "architectures": [ + "GPT2LMHeadModel" + ], + "attn_mult": 8.0, + "attn_pdrop": 0.1, + "bos_token_id": 50256, + "embd_pdrop": 0.1, + "eos_token_id": 50256, + "initializer_range": 0.01, + "intermediate_size": 8192, + "layer_norm_epsilon": 1e-05, + "model_type": "gpt2", + "n_embd": 2048, + "n_head": 64, + "n_inner": null, + "n_layer": 12, + "n_positions": 1024, + "num_layers": 12, + "reorder_and_upcast_attn": false, + "resid_pdrop": 0.1, + "scale_attn_by_inverse_layer_idx": false, + "scale_attn_weights": true, + "summary_activation": null, + "summary_first_dropout": 0.1, + "summary_proj_to_labels": true, + "summary_type": "cls_index", + "summary_use_proj": true, + "torch_dtype": "float32", + "transformers_version": "4.25.1", + "use_cache": false, + "vocab_size": 50257 +} diff --git a/mup-800m-100m/checkpoint-200/optimizer.pt b/mup-800m-100m/checkpoint-200/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..bef41093826372e87e972bdfc8377c6bd7766392 --- /dev/null +++ b/mup-800m-100m/checkpoint-200/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c60ec89bab2ed93738d18681521a8325a9b3b847e0d5869a0dc755c6de7a5240 +size 5674738681 diff --git a/mup-800m-100m/checkpoint-200/pytorch_model.bin b/mup-800m-100m/checkpoint-200/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..139dc39a619964463a7dc404e0d35cc6f86e790a --- /dev/null +++ b/mup-800m-100m/checkpoint-200/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a6d85a2f36c665ace27b6a251aed46c9eeac1dc3b1b076f4a62f63178d1e38c +size 2849946173 diff --git a/mup-800m-100m/checkpoint-200/rng_state.pth b/mup-800m-100m/checkpoint-200/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..caecf4284bfd694aaba3fff8c317c3fca9f5411a --- /dev/null +++ b/mup-800m-100m/checkpoint-200/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a02e73f7c8974df633ab86c99ffd01ab988a956ee7165e74c535f0f9be5364f0 +size 14575 diff --git a/mup-800m-100m/checkpoint-200/scaler.pt b/mup-800m-100m/checkpoint-200/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..87710d0ddc627c070366fcb3112b07dc60d97295 --- /dev/null +++ b/mup-800m-100m/checkpoint-200/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4fccf0f9be1bb8f24861e4393745b3e09cc2687125a69e3757955fb0f0925ea5 +size 557 diff --git a/mup-800m-100m/checkpoint-200/scheduler.pt b/mup-800m-100m/checkpoint-200/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..9858e5ec85a140ecccf75019fa3976cb82ae06f8 --- /dev/null +++ b/mup-800m-100m/checkpoint-200/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f6e4d2b46fef519f0cc5111d11c4ebbe6d88971edc8739328809db9f143c06e +size 691 diff --git a/mup-800m-100m/checkpoint-200/trainer_state.json b/mup-800m-100m/checkpoint-200/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..875429697fde116b63adbe0e3d8bb945bd35cf2a --- /dev/null +++ b/mup-800m-100m/checkpoint-200/trainer_state.json @@ -0,0 +1,15 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.5261967893774023, + "global_step": 200, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [], + "max_steps": 380, + "num_train_epochs": 1, + "total_flos": 1.90097400004608e+17, + "trial_name": null, + "trial_params": null +} diff --git a/mup-800m-100m/checkpoint-200/training_args.bin b/mup-800m-100m/checkpoint-200/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..507802263d44f3aa55e5dba70d42b153113db0d9 --- /dev/null +++ b/mup-800m-100m/checkpoint-200/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e10f7310880f2f2fc7fc715f393057fe3a9edc2aef2788ec38f44a4faa96083 +size 3387 diff --git a/mup-800m-100m/checkpoint-300/config.json b/mup-800m-100m/checkpoint-300/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cd4fa9efea20031e766b401a8f2b6481d482fde0 --- /dev/null +++ b/mup-800m-100m/checkpoint-300/config.json @@ -0,0 +1,34 @@ +{ + "activation_function": "gelu_new", + "architectures": [ + "GPT2LMHeadModel" + ], + "attn_mult": 8.0, + "attn_pdrop": 0.1, + "bos_token_id": 50256, + "embd_pdrop": 0.1, + "eos_token_id": 50256, + "initializer_range": 0.01, + "intermediate_size": 8192, + "layer_norm_epsilon": 1e-05, + "model_type": "gpt2", + "n_embd": 2048, + "n_head": 64, + "n_inner": null, + "n_layer": 12, + "n_positions": 1024, + "num_layers": 12, + "reorder_and_upcast_attn": false, + "resid_pdrop": 0.1, + "scale_attn_by_inverse_layer_idx": false, + "scale_attn_weights": true, + "summary_activation": null, + "summary_first_dropout": 0.1, + "summary_proj_to_labels": true, + "summary_type": "cls_index", + "summary_use_proj": true, + "torch_dtype": "float32", + "transformers_version": "4.25.1", + "use_cache": false, + "vocab_size": 50257 +} diff --git a/mup-800m-100m/checkpoint-300/optimizer.pt b/mup-800m-100m/checkpoint-300/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..05ad8d7a6ba9f4bcf7f0d527eebb936f5d690ec2 --- /dev/null +++ b/mup-800m-100m/checkpoint-300/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa43e73f27ce23f9e6ac25cc9d626ae1df742951d3113ac603481f6276124afd +size 5674738681 diff --git a/mup-800m-100m/checkpoint-300/pytorch_model.bin b/mup-800m-100m/checkpoint-300/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..9b56e1d653b78d9612846abb4d0511c46a52083a --- /dev/null +++ b/mup-800m-100m/checkpoint-300/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8568942a9cd4d7a58f95641d13f6cce64a54afd213a8d730af616b62579581a +size 2849946173 diff --git a/mup-800m-100m/checkpoint-300/rng_state.pth b/mup-800m-100m/checkpoint-300/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..3f0c712fdc22e8cbdc4da49611638946644e77d2 --- /dev/null +++ b/mup-800m-100m/checkpoint-300/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9b81be87d3a4a0333c3e9463067829ade4eff5466bc50d43f413021e46b0083 +size 14575 diff --git a/mup-800m-100m/checkpoint-300/scaler.pt b/mup-800m-100m/checkpoint-300/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..96d1b8ea2cbad321ffe9c10840a99bb1bcef18f5 --- /dev/null +++ b/mup-800m-100m/checkpoint-300/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efafd90182e3d39d1b7c4a686f86e5913f5abc094dc3e2f827a6d479c6cef247 +size 557 diff --git a/mup-800m-100m/checkpoint-300/scheduler.pt b/mup-800m-100m/checkpoint-300/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..d6553df45750cc7e0c10be0b79f74d283c06dfcf --- /dev/null +++ b/mup-800m-100m/checkpoint-300/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:202d672dbe4a889df878d8f3232ede044f65c23bdde3de48edafae6f2790cace +size 691 diff --git a/mup-800m-100m/checkpoint-300/trainer_state.json b/mup-800m-100m/checkpoint-300/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..e6edae1f95c6847bbde2c12b83055be42671eb67 --- /dev/null +++ b/mup-800m-100m/checkpoint-300/trainer_state.json @@ -0,0 +1,15 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.7892951840661034, + "global_step": 300, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [], + "max_steps": 380, + "num_train_epochs": 1, + "total_flos": 2.85146100006912e+17, + "trial_name": null, + "trial_params": null +} diff --git a/mup-800m-100m/checkpoint-300/training_args.bin b/mup-800m-100m/checkpoint-300/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..507802263d44f3aa55e5dba70d42b153113db0d9 --- /dev/null +++ b/mup-800m-100m/checkpoint-300/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e10f7310880f2f2fc7fc715f393057fe3a9edc2aef2788ec38f44a4faa96083 +size 3387 diff --git a/mup-800m-100m/config.json b/mup-800m-100m/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cd4fa9efea20031e766b401a8f2b6481d482fde0 --- /dev/null +++ b/mup-800m-100m/config.json @@ -0,0 +1,34 @@ +{ + "activation_function": "gelu_new", + "architectures": [ + "GPT2LMHeadModel" + ], + "attn_mult": 8.0, + "attn_pdrop": 0.1, + "bos_token_id": 50256, + "embd_pdrop": 0.1, + "eos_token_id": 50256, + "initializer_range": 0.01, + "intermediate_size": 8192, + "layer_norm_epsilon": 1e-05, + "model_type": "gpt2", + "n_embd": 2048, + "n_head": 64, + "n_inner": null, + "n_layer": 12, + "n_positions": 1024, + "num_layers": 12, + "reorder_and_upcast_attn": false, + "resid_pdrop": 0.1, + "scale_attn_by_inverse_layer_idx": false, + "scale_attn_weights": true, + "summary_activation": null, + "summary_first_dropout": 0.1, + "summary_proj_to_labels": true, + "summary_type": "cls_index", + "summary_use_proj": true, + "torch_dtype": "float32", + "transformers_version": "4.25.1", + "use_cache": false, + "vocab_size": 50257 +} diff --git a/mup-800m-100m/pytorch_model.bin b/mup-800m-100m/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..2040c263107e38fcd3fc93cbb7f32d316e1ab565 --- /dev/null +++ b/mup-800m-100m/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9b1e50db60ab168a8a366abdb75b27144fc8e7e4fb44b503bfe9f3243472983 +size 2849946173 diff --git a/mup-800m-100m/training_args.bin b/mup-800m-100m/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..507802263d44f3aa55e5dba70d42b153113db0d9 --- /dev/null +++ b/mup-800m-100m/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e10f7310880f2f2fc7fc715f393057fe3a9edc2aef2788ec38f44a4faa96083 +size 3387 diff --git a/mup-2b-100m-e3/config.json b/sp-2b-100m/checkpoint-300/config.json similarity index 91% rename from mup-2b-100m-e3/config.json rename to sp-2b-100m/checkpoint-300/config.json index 206494207d182569dd9eedb97a900ced82bdf150..2de73dfd00a43b5790e279806c895c79ddedd3e4 100644 --- a/mup-2b-100m-e3/config.json +++ b/sp-2b-100m/checkpoint-300/config.json @@ -3,12 +3,11 @@ "architectures": [ "GPT2LMHeadModel" ], - "attn_mult": 8.0, "attn_pdrop": 0.1, "bos_token_id": 50256, "embd_pdrop": 0.1, "eos_token_id": 50256, - "initializer_range": 0.01, + "initializer_range": 0.02, "intermediate_size": 13312, "layer_norm_epsilon": 1e-05, "model_type": "gpt2", @@ -29,6 +28,6 @@ "summary_use_proj": true, "torch_dtype": "float32", "transformers_version": "4.25.1", - "use_cache": true, + "use_cache": false, "vocab_size": 50257 } diff --git a/sp-2b-100m/checkpoint-300/optimizer.pt b/sp-2b-100m/checkpoint-300/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..1bd32590dd5b73e1d86c7e75eefe518a7777fe3b --- /dev/null +++ b/sp-2b-100m/checkpoint-300/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c94bef3fa838e5fd53131c9c6898011e9bdd37373dc68d0b2daf6736b3bdd5b +size 14128673937 diff --git a/sp-2b-100m/checkpoint-300/pytorch_model.bin b/sp-2b-100m/checkpoint-300/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..bded2b9d19ff693c7dff059d85c37f3cc1772716 --- /dev/null +++ b/sp-2b-100m/checkpoint-300/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3542d421d3ff0fd029545b3e74fdd3664bc977587f009a84629ebad867c7821e +size 7076932005 diff --git a/mup-2b-100m-e3/checkpoint-300/rng_state.pth b/sp-2b-100m/checkpoint-300/rng_state.pth similarity index 100% rename from mup-2b-100m-e3/checkpoint-300/rng_state.pth rename to sp-2b-100m/checkpoint-300/rng_state.pth diff --git a/sp-2b-100m/checkpoint-300/scaler.pt b/sp-2b-100m/checkpoint-300/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..96d1b8ea2cbad321ffe9c10840a99bb1bcef18f5 --- /dev/null +++ b/sp-2b-100m/checkpoint-300/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efafd90182e3d39d1b7c4a686f86e5913f5abc094dc3e2f827a6d479c6cef247 +size 557 diff --git a/sp-2b-100m/checkpoint-300/scheduler.pt b/sp-2b-100m/checkpoint-300/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..89a3685511ff0baf0caa02c2ab8f23fa6e214317 --- /dev/null +++ b/sp-2b-100m/checkpoint-300/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4546d9532f4bee54bea33da91e5c41d94c56696c6723d0717ee8575a4c90c582 +size 627 diff --git a/mup-2b-100m-e3/checkpoint-300/trainer_state.json b/sp-2b-100m/checkpoint-300/trainer_state.json similarity index 100% rename from mup-2b-100m-e3/checkpoint-300/trainer_state.json rename to sp-2b-100m/checkpoint-300/trainer_state.json diff --git a/sp-2b-100m/checkpoint-300/training_args.bin b/sp-2b-100m/checkpoint-300/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..f9d45fb602fffe4cb0d925595582c2194cb195e3 --- /dev/null +++ b/sp-2b-100m/checkpoint-300/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69bb078ba439c49ec9bf40cd7f966b73c70dfe5796ee2bb289768a61d58d7ff2 +size 3387 diff --git a/mup-2b-100m-e3/checkpoint-100/config.json b/sp-2b-100m/config.json similarity index 91% rename from mup-2b-100m-e3/checkpoint-100/config.json rename to sp-2b-100m/config.json index 206494207d182569dd9eedb97a900ced82bdf150..2de73dfd00a43b5790e279806c895c79ddedd3e4 100644 --- a/mup-2b-100m-e3/checkpoint-100/config.json +++ b/sp-2b-100m/config.json @@ -3,12 +3,11 @@ "architectures": [ "GPT2LMHeadModel" ], - "attn_mult": 8.0, "attn_pdrop": 0.1, "bos_token_id": 50256, "embd_pdrop": 0.1, "eos_token_id": 50256, - "initializer_range": 0.01, + "initializer_range": 0.02, "intermediate_size": 13312, "layer_norm_epsilon": 1e-05, "model_type": "gpt2", @@ -29,6 +28,6 @@ "summary_use_proj": true, "torch_dtype": "float32", "transformers_version": "4.25.1", - "use_cache": true, + "use_cache": false, "vocab_size": 50257 } diff --git a/sp-2b-100m/pytorch_model.bin b/sp-2b-100m/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..059837a0a009870c1aadf0cfaa97db3e04e089cc --- /dev/null +++ b/sp-2b-100m/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:adcbf18eda2460dbc1997f577acd1f4c6f38d942077556423928369f7c45e7ee +size 7076932005 diff --git a/sp-2b-100m/training_args.bin b/sp-2b-100m/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..f9d45fb602fffe4cb0d925595582c2194cb195e3 --- /dev/null +++ b/sp-2b-100m/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69bb078ba439c49ec9bf40cd7f966b73c70dfe5796ee2bb289768a61d58d7ff2 +size 3387 diff --git a/sp-2b5-100m/config.json b/sp-2b5-100m/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f1e75fadc018ee6ef27901e1ac36668c3578726b --- /dev/null +++ b/sp-2b5-100m/config.json @@ -0,0 +1,33 @@ +{ + "activation_function": "gelu_new", + "architectures": [ + "GPT2LMHeadModel" + ], + "attn_pdrop": 0.1, + "bos_token_id": 50256, + "embd_pdrop": 0.1, + "eos_token_id": 50256, + "initializer_range": 0.02, + "intermediate_size": 16384, + "layer_norm_epsilon": 1e-05, + "model_type": "gpt2", + "n_embd": 4096, + "n_head": 128, + "n_inner": null, + "n_layer": 12, + "n_positions": 1024, + "num_layers": 12, + "reorder_and_upcast_attn": false, + "resid_pdrop": 0.1, + "scale_attn_by_inverse_layer_idx": false, + "scale_attn_weights": true, + "summary_activation": null, + "summary_first_dropout": 0.1, + "summary_proj_to_labels": true, + "summary_type": "cls_index", + "summary_use_proj": true, + "torch_dtype": "float32", + "transformers_version": "4.25.1", + "use_cache": false, + "vocab_size": 50257 +} diff --git a/sp-2b5-100m/pytorch_model-00001-of-00002.bin b/sp-2b5-100m/pytorch_model-00001-of-00002.bin new file mode 100644 index 0000000000000000000000000000000000000000..750b876110c932552ee58a9c66c7b80e7c3ec6fc --- /dev/null +++ b/sp-2b5-100m/pytorch_model-00001-of-00002.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c61766cf18dc6fb818d550af5c8502d17ace3a4a8d61c62040af01f42245954f +size 9982107965 diff --git a/sp-2b5-100m/pytorch_model-00002-of-00002.bin b/sp-2b5-100m/pytorch_model-00002-of-00002.bin new file mode 100644 index 0000000000000000000000000000000000000000..ed46721743343ad4d6cdb79f3a3e6c009793f21e --- /dev/null +++ b/sp-2b5-100m/pytorch_model-00002-of-00002.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2207816e703cdf835c8a89c7856fb7909a1baeb08283b264dc7514ee0e2933a +size 1360399056 diff --git a/mup-3b-100m-e3/pytorch_model.bin.index.json b/sp-2b5-100m/pytorch_model.bin.index.json similarity index 89% rename from mup-3b-100m-e3/pytorch_model.bin.index.json rename to sp-2b5-100m/pytorch_model.bin.index.json index c25d2ada1b3f0909e5b391003284951a36fa6a5a..d31972f5835b2ee854ba8dba6ecb387529392231 100644 --- a/mup-3b-100m-e3/pytorch_model.bin.index.json +++ b/sp-2b5-100m/pytorch_model.bin.index.json @@ -1,6 +1,6 @@ { "metadata": { - "total_size": 12692291632 + "total_size": 11342446640 }, "weight_map": { "lm_head.weight": "pytorch_model-00002-of-00002.bin", @@ -33,29 +33,29 @@ "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", "transformer.h.10.attn.bias": "pytorch_model-00001-of-00002.bin", - "transformer.h.10.attn.c_attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.c_attn.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.attn.c_proj.weight": "pytorch_model-00002-of-00002.bin", + "transformer.h.10.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", "transformer.h.10.attn.masked_bias": "pytorch_model-00001-of-00002.bin", "transformer.h.10.ln_1.bias": "pytorch_model-00001-of-00002.bin", "transformer.h.10.ln_1.weight": "pytorch_model-00001-of-00002.bin", - "transformer.h.10.ln_2.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.ln_2.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_attn.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_attn.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_proj.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.c_proj.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.attn.masked_bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_1.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_1.weight": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_2.bias": "pytorch_model-00002-of-00002.bin", - "transformer.h.11.ln_2.weight": "pytorch_model-00002-of-00002.bin", + "transformer.h.10.ln_2.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.ln_2.weight": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.attn.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.attn.c_attn.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.attn.c_attn.weight": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.attn.c_proj.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.attn.c_proj.weight": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.attn.masked_bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.ln_1.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.ln_1.weight": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.ln_2.bias": "pytorch_model-00001-of-00002.bin", + "transformer.h.11.ln_2.weight": "pytorch_model-00001-of-00002.bin", "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00002-of-00002.bin", "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00002-of-00002.bin", "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00002-of-00002.bin", diff --git a/sp-2b5-100m/training_args.bin b/sp-2b5-100m/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..38d75dc6e1242b19501f21818d31988036c6647b --- /dev/null +++ b/sp-2b5-100m/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f140419ae8e7b1ae56b2a370dbc805cd034c9bb80725c93e6d257c60a394fe8 +size 3387 diff --git a/sp-3b5-100m-e3-full.json b/sp-3b5-100m-e3-full.json deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000