diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000000000000000000000000000000000000..a6344aac8c09253b3b630fb776ae94478aa0275b --- /dev/null +++ b/.gitattributes @@ -0,0 +1,35 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..e2ecde6546a1bb47171ca826ab969d6be7c26fb6 --- /dev/null +++ b/README.md @@ -0,0 +1,33 @@ +--- +license: mit +tags: +- reasoning +- puzzles +- sudoku +- maze +- test-time-search +- tiny-recursive-models +--- + +# ICS checkpoints + +Checkpoints for ICS (Input-Conditioned Search; code: https://github.com/hcshi316/ICS) and the baselines. + +## Layout + +`//seed/` holds `config.json` and `model.safetensors`. + +## Loading + +```python +from huggingface_hub import snapshot_download + +root = snapshot_download(repo_id="hcshi/ICS", allow_patterns="trm/maze/seed0/*") +``` + +With the code installed, `python -m ics download --model trm --task maze --seed 0` writes the same folder to +`ckpt/trm/maze/seed0`. + +## License + +MIT. diff --git a/attractor/heyawake/seed0/config.json b/attractor/heyawake/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..db868eb619698ac38a1ebde7b104a0538eda2ba0 --- /dev/null +++ b/attractor/heyawake/seed0/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/heyawake/seed0/model.safetensors b/attractor/heyawake/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fa79ff1305826ed505961e719d8b0a702e695bb1 --- /dev/null +++ b/attractor/heyawake/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc960d6157d751c52c1e3cd90ddf2b7fae66850dbfddfa3f4cdef53bffe52c69 +size 110678368 diff --git a/attractor/lightup/seed0/config.json b/attractor/lightup/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a9e00cfd51791955e87f865beabfabd5de94e067 --- /dev/null +++ b/attractor/lightup/seed0/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/lightup/seed0/model.safetensors b/attractor/lightup/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ac26551f6394bb1061fea595b3f3f5b0e69dfe92 --- /dev/null +++ b/attractor/lightup/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f967b1b6a58a731ac0403adc678a85342e014ae8f43c9bf23e9a89cd2f8673f +size 109101400 diff --git a/attractor/maze/seed0/config.json b/attractor/maze/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b918fd850d7776d404f54bff75ffb99390dc5804 --- /dev/null +++ b/attractor/maze/seed0/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/maze/seed0/model.safetensors b/attractor/maze/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fbb91601d1505ec3d68bc9b7d1b2590cc3dd1bc3 --- /dev/null +++ b/attractor/maze/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e54c7de109b7e155277910de3ce4b2170084be7fe5b77d852430b8dca356c72 +size 109089112 diff --git a/attractor/nurikabe/seed0/config.json b/attractor/nurikabe/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..22f365ad49d0733f4139e240aba1e7c0df427a34 --- /dev/null +++ b/attractor/nurikabe/seed0/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/nurikabe/seed0/model.safetensors b/attractor/nurikabe/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..83ac1c8fddcad8566bc6748cc874210a304abfeb --- /dev/null +++ b/attractor/nurikabe/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ccddf1c3574191d8e122a20b3d9cb2d9db465003e30d4ae7f698643dc0020c8 +size 109293920 diff --git a/attractor/sudoku/seed0/config.json b/attractor/sudoku/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2cbe2e240f2f113ec182741b3c282a2326f36284 --- /dev/null +++ b/attractor/sudoku/seed0/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/sudoku/seed0/model.safetensors b/attractor/sudoku/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c56c08badd6a1dbfc21a6349ad86b42d65924d1f --- /dev/null +++ b/attractor/sudoku/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70535c0d8f26e35265c71c4796a4766aa135e6a6aee2996ac7bd337466075c09 +size 109109600 diff --git a/attractor/sudoku/seed1/config.json b/attractor/sudoku/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2cbe2e240f2f113ec182741b3c282a2326f36284 --- /dev/null +++ b/attractor/sudoku/seed1/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/sudoku/seed1/model.safetensors b/attractor/sudoku/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b07e2cacd1a9fa2925c55bc1b59804de7a79287a --- /dev/null +++ b/attractor/sudoku/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16fd81bd24f450e9b6a74f0c2f1c62e2e1fd1e84500d7b45fe09918fb9df644f +size 109109600 diff --git a/attractor/sudoku/seed2/config.json b/attractor/sudoku/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2cbe2e240f2f113ec182741b3c282a2326f36284 --- /dev/null +++ b/attractor/sudoku/seed2/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/sudoku/seed2/model.safetensors b/attractor/sudoku/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..48433f680525d710101b41b95bae3bdfd5af3fe5 --- /dev/null +++ b/attractor/sudoku/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b954f13131a2045c33412ee4b534fcf58b6566b64694971df1a0f40ec2b9c11e +size 109109600 diff --git a/attractor/tapa/seed0/config.json b/attractor/tapa/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f4437fa619908aa55107c4bf79908fa4aa722c8e --- /dev/null +++ b/attractor/tapa/seed0/config.json @@ -0,0 +1,31 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 8, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 96, + "deq_max_iter": 8, + "deq_min_iter": 4, + "deq_tol": 0.001, + "deq_anderson_m": 5, + "deq_anderson_beta": 1.0, + "bptt_through": 2, + "jacobian_reg_lambda": 0.001 + }, + "provenance": {} +} \ No newline at end of file diff --git a/attractor/tapa/seed0/model.safetensors b/attractor/tapa/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6e338f9e099ed84c2d8247d342e3165598b6eca6 --- /dev/null +++ b/attractor/tapa/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57f699f12a42bf54e6e85013c3239f9d838ff0077640c6005923c5487217689d +size 109232480 diff --git a/eqr/heyawake/seed0/config.json b/eqr/heyawake/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4cb9eb1285c72ba7c36d0dc8be1f855b89440317 --- /dev/null +++ b/eqr/heyawake/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": true, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/heyawake/seed0/model.safetensors b/eqr/heyawake/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ec90f52653212acb2b3933a3b16df9ae703bca43 --- /dev/null +++ b/eqr/heyawake/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6815fc8adade187d77431ea8fc089cde4be9378054ccc8cab3c68c1eea5bf8ca +size 53720368 diff --git a/eqr/lightup/seed0/config.json b/eqr/lightup/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..078e6affa348acaad93a0f0771c6cc26da8a90f7 --- /dev/null +++ b/eqr/lightup/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": true, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/lightup/seed0/model.safetensors b/eqr/lightup/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ab540b89bc3db278b85f22c845e84816dfe0dc10 --- /dev/null +++ b/eqr/lightup/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a028cfb0b2297f4ecad8c559a2492df4d9b11a1f657205f9d18a8a871ae8ef18 +size 52143408 diff --git a/eqr/maze/seed0/config.json b/eqr/maze/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..14a8716fa359fab302a4a36936fc4803758e9172 --- /dev/null +++ b/eqr/maze/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/maze/seed0/model.safetensors b/eqr/maze/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f8b169fadbb9f628d246afa37e959a130f328f3b --- /dev/null +++ b/eqr/maze/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9499b3f9bf5150910fb991dd3442142e4c44e4e6d3c7b3dabd03d88a2eef01fc +size 27292976 diff --git a/eqr/maze/seed1/config.json b/eqr/maze/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..14a8716fa359fab302a4a36936fc4803758e9172 --- /dev/null +++ b/eqr/maze/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/maze/seed1/model.safetensors b/eqr/maze/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..223dd9d501dfdce33072ea019d715ee615838736 --- /dev/null +++ b/eqr/maze/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eebf12af78843df720329fb52da57c4646ebb8e251c486b6b93b0387ad36fdd2 +size 27292976 diff --git a/eqr/maze/seed2/config.json b/eqr/maze/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..14a8716fa359fab302a4a36936fc4803758e9172 --- /dev/null +++ b/eqr/maze/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/maze/seed2/model.safetensors b/eqr/maze/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a18bed58c43b6968c083adc304368e32f43f7c31 --- /dev/null +++ b/eqr/maze/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26ecd1f8fbb49d94b204ec81928a57e41965424517c6c3fb90d34cd9df8889d9 +size 27292976 diff --git a/eqr/nurikabe/seed0/config.json b/eqr/nurikabe/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fda623420458f04bd89ff1fc4851cac6ec342bd7 --- /dev/null +++ b/eqr/nurikabe/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": true, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/nurikabe/seed0/model.safetensors b/eqr/nurikabe/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ae97f74e6f1dad05bfc03f491f60fe970c8e3771 --- /dev/null +++ b/eqr/nurikabe/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2b839d5b4826cc78d685d67cd6a5574734af72192186efb762f94b7462a8f10 +size 52335920 diff --git a/eqr/sudoku/seed0/config.json b/eqr/sudoku/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3adf831cc8e4ee707c33466d75f306be5fa779fb --- /dev/null +++ b/eqr/sudoku/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": true, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/sudoku/seed0/model.safetensors b/eqr/sudoku/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f54c244bc5f22022ffbd33c693f0f07a7246fab7 --- /dev/null +++ b/eqr/sudoku/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e278f13f8048266e6c5946f69e6fd6ee44768c71a45a17958183ca94dd6412ae +size 19422496 diff --git a/eqr/sudoku/seed1/config.json b/eqr/sudoku/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3adf831cc8e4ee707c33466d75f306be5fa779fb --- /dev/null +++ b/eqr/sudoku/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": true, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/sudoku/seed1/model.safetensors b/eqr/sudoku/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..efd94b6a798dd4573f90ecf4b3a459e974d2c951 --- /dev/null +++ b/eqr/sudoku/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aefa6a304184ec18f91b83947b117d0bb021df90430b94094a735728ef7ab187 +size 19422496 diff --git a/eqr/sudoku/seed2/config.json b/eqr/sudoku/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3adf831cc8e4ee707c33466d75f306be5fa779fb --- /dev/null +++ b/eqr/sudoku/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": true, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/sudoku/seed2/model.safetensors b/eqr/sudoku/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8550a18e5a0d75e7731a4b9fb2fd446b30d01e0d --- /dev/null +++ b/eqr/sudoku/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4ba1aa981bcb3caac88a312887602ac0457db604d42171057cd2e7673adb4f4 +size 19422496 diff --git a/eqr/tapa/seed0/config.json b/eqr/tapa/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3ed9ed653d842c42c08c9e11b7738b38e17d5a88 --- /dev/null +++ b/eqr/tapa/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": true, + "lambda_": 0.95, + "noise_scale": 0.01, + "H_init_std": 1.0, + "L_init_std": 1.0, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/eqr/tapa/seed0/model.safetensors b/eqr/tapa/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2995475ac0e0b1c2d91cb5b67fe748738de5d09f --- /dev/null +++ b/eqr/tapa/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0d12a9c4f529de509da787fe6efb28ad3b0a8673fcc312916dae7d6857085d3 +size 52274480 diff --git a/gram/heyawake/seed0/config.json b/gram/heyawake/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3c7d1da416d4e651a88db1889517139462aaff12 --- /dev/null +++ b/gram/heyawake/seed0/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/heyawake/seed0/model.safetensors b/gram/heyawake/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9534d6b0e06487159fe05aee3ceff9a4112a97c7 --- /dev/null +++ b/gram/heyawake/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:55a85a8bacdd915b6d8257d85c81eddaa6d881a688b9e4e76501af65706b83ee +size 69815324 diff --git a/gram/heyawake/seed1/config.json b/gram/heyawake/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3c7d1da416d4e651a88db1889517139462aaff12 --- /dev/null +++ b/gram/heyawake/seed1/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/heyawake/seed1/model.safetensors b/gram/heyawake/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..21c9c4f3bfb3ea5ed629bac8f7a8b54f4799f5ed --- /dev/null +++ b/gram/heyawake/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b36499e7332228e64b0281a0b9c59aeee3c30db1f28da26383ff5c35d0252c9 +size 69815324 diff --git a/gram/heyawake/seed2/config.json b/gram/heyawake/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3c7d1da416d4e651a88db1889517139462aaff12 --- /dev/null +++ b/gram/heyawake/seed2/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/heyawake/seed2/model.safetensors b/gram/heyawake/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5dff31f42c2833122fd2038709127164e7c42b15 --- /dev/null +++ b/gram/heyawake/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8bd5a65a07bcd0f7a48c0d6e058f006b27d39540d08eb57bd9a7d14a0d0edddd +size 69815324 diff --git a/gram/lightup/seed0/config.json b/gram/lightup/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..253140cd0d815a7b3171b7123fa83b2b19dc04e7 --- /dev/null +++ b/gram/lightup/seed0/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/lightup/seed0/model.safetensors b/gram/lightup/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c8531fbb6f0e55ccd0d33674091335147ae4f17c --- /dev/null +++ b/gram/lightup/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c4458c7d120ba609d77005cd669c18a7623398014c68a2c721948b50568b211 +size 68238356 diff --git a/gram/lightup/seed1/config.json b/gram/lightup/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..253140cd0d815a7b3171b7123fa83b2b19dc04e7 --- /dev/null +++ b/gram/lightup/seed1/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/lightup/seed1/model.safetensors b/gram/lightup/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7844a267194ffa0d22ef89ff2eebc3a8471f1b05 --- /dev/null +++ b/gram/lightup/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c36fe3ce43407857ad9a74479d3f596e0e63b33a5b4eb40196faa918357685b5 +size 68238356 diff --git a/gram/lightup/seed2/config.json b/gram/lightup/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..253140cd0d815a7b3171b7123fa83b2b19dc04e7 --- /dev/null +++ b/gram/lightup/seed2/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/lightup/seed2/model.safetensors b/gram/lightup/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9fe97c3d5367da88b3785ab33b2e79542f7a94e9 --- /dev/null +++ b/gram/lightup/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0cce6a22720da1220ff648bd7e9255ab36ccb0f1970f094c67be7523c1aa413 +size 68238356 diff --git a/gram/maze/seed0/config.json b/gram/maze/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5e1225c98da75939be55e7f3ace364ceddc53f30 --- /dev/null +++ b/gram/maze/seed0/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/maze/seed0/model.safetensors b/gram/maze/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7468737fe508bd33dced9e0275922dd6529c8c83 --- /dev/null +++ b/gram/maze/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:312d6a5e84566b646db13f682995a6ad469b5b42dfb05f41b99709c77682c15c +size 68226068 diff --git a/gram/maze/seed1/config.json b/gram/maze/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5e1225c98da75939be55e7f3ace364ceddc53f30 --- /dev/null +++ b/gram/maze/seed1/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/maze/seed1/model.safetensors b/gram/maze/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7ceba0ca36f776b42ce03f14d91ce4a2f096aaa4 --- /dev/null +++ b/gram/maze/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:24b2af53e9834af2e897f3f3911ddc4f2d5eac2c5b59da11278338a334389ad7 +size 68226068 diff --git a/gram/maze/seed2/config.json b/gram/maze/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5e1225c98da75939be55e7f3ace364ceddc53f30 --- /dev/null +++ b/gram/maze/seed2/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/maze/seed2/model.safetensors b/gram/maze/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..dde12399d0406ac7b1348f4aa28d1d557454dfcb --- /dev/null +++ b/gram/maze/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa9b989c2d0fdab7fdffd63c9a776764a3b971b5c223fada020dee964d7a9a08 +size 68226068 diff --git a/gram/nurikabe/seed0/config.json b/gram/nurikabe/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2d79e4d5f92cbf71b9d75f67e83aec7cc427309d --- /dev/null +++ b/gram/nurikabe/seed0/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/nurikabe/seed0/model.safetensors b/gram/nurikabe/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c22c6de8137abd2272b6dfa7ff8e6a7832c86c98 --- /dev/null +++ b/gram/nurikabe/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41faaab16b739290bbc79223e09dd41a4e942053ccb4982e2189f40e123b33dd +size 68430868 diff --git a/gram/nurikabe/seed1/config.json b/gram/nurikabe/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2d79e4d5f92cbf71b9d75f67e83aec7cc427309d --- /dev/null +++ b/gram/nurikabe/seed1/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/nurikabe/seed1/model.safetensors b/gram/nurikabe/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f0061bb7e39faf2ea969e3ae6b2320c480cda32b --- /dev/null +++ b/gram/nurikabe/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:902eb47d831ec633626fe6f53bc5041f11cd7a4b154918512027e27068fa652b +size 68430868 diff --git a/gram/nurikabe/seed2/config.json b/gram/nurikabe/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2d79e4d5f92cbf71b9d75f67e83aec7cc427309d --- /dev/null +++ b/gram/nurikabe/seed2/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/nurikabe/seed2/model.safetensors b/gram/nurikabe/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..30c6c3ec5d65364e7b87b1d15334b3c1fa9554a2 --- /dev/null +++ b/gram/nurikabe/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae6a704017ceb91529dc48f6ff401cbcc1538ad14c6a6bff1f65205419f9876c +size 68430868 diff --git a/gram/sudoku/seed0/config.json b/gram/sudoku/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7170f5cdd1359edc3139968ce4c60c21af8ca89c --- /dev/null +++ b/gram/sudoku/seed0/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "none", + "mlp_t": true, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/sudoku/seed0/model.safetensors b/gram/sudoku/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bceae3fe1541e6352dda7c8c292b942dfe41113e --- /dev/null +++ b/gram/sudoku/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59e406bf84d2448bb7f2ff5b127c3100a508e55520faf7e4fcb41f4e88eff532 +size 53853188 diff --git a/gram/sudoku/seed1/config.json b/gram/sudoku/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7170f5cdd1359edc3139968ce4c60c21af8ca89c --- /dev/null +++ b/gram/sudoku/seed1/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "none", + "mlp_t": true, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/sudoku/seed1/model.safetensors b/gram/sudoku/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..147b7bb680b4ab2c159e656fb8dc74165206b541 --- /dev/null +++ b/gram/sudoku/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f4798a6c5d44a2d88c4b03638b2e0cba4f9089f020a5bdad76f06dff888329a +size 53853188 diff --git a/gram/sudoku/seed2/config.json b/gram/sudoku/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7170f5cdd1359edc3139968ce4c60c21af8ca89c --- /dev/null +++ b/gram/sudoku/seed2/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "none", + "mlp_t": true, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/sudoku/seed2/model.safetensors b/gram/sudoku/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c19f44b8f556b0e47006c53910c3e44207cf985c --- /dev/null +++ b/gram/sudoku/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b69e7a45235cf45088ccc8469df8d62b42e680382dbe36477959c999f9aade9a +size 53853188 diff --git a/gram/tapa/seed0/config.json b/gram/tapa/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a304b694f4e13707b912fe33a9159585299ae368 --- /dev/null +++ b/gram/tapa/seed0/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/tapa/seed0/model.safetensors b/gram/tapa/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e9e1e833fab852efaab22a18b60498746dce5cdd --- /dev/null +++ b/gram/tapa/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a26a7635007a8d5b7e6d793418ce6aa5ceb255c396c2cc87afe14e91eb52dd9 +size 68369428 diff --git a/gram/tapa/seed1/config.json b/gram/tapa/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a304b694f4e13707b912fe33a9159585299ae368 --- /dev/null +++ b/gram/tapa/seed1/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/tapa/seed1/model.safetensors b/gram/tapa/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4e15361064d4ba5424f711caee88ef0a6f7cf51c --- /dev/null +++ b/gram/tapa/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e072e92b87b0ed841aa8b8ae863e5c8f44673367c07bc32fa968ee68b9ed86f +size 68369428 diff --git a/gram/tapa/seed2/config.json b/gram/tapa/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a304b694f4e13707b912fe33a9159585299ae368 --- /dev/null +++ b/gram/tapa/seed2/config.json @@ -0,0 +1,27 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "num_puzzle_identifiers": 1, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "noise_expansion": 1.0, + "min_std": 0.001, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "kl_coef": 0.1, + "kl_balance": 0.8, + "kl_all_transitions": true, + "grad_checkpoint": true, + "forward_dtype": "bfloat16" + }, + "provenance": {} +} \ No newline at end of file diff --git a/gram/tapa/seed2/model.safetensors b/gram/tapa/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9c3d37368b3f788768ca0e5f2cdb38337c562bab --- /dev/null +++ b/gram/tapa/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44ac6fc4750a995c104eeb7c7faf2574cf9deff5e57b9de6872a62e883228dd0 +size 68369428 diff --git a/trm/heyawake/seed0/config.json b/trm/heyawake/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d4a8c72610623ed37707d16b0d617a3a3b7ebd21 --- /dev/null +++ b/trm/heyawake/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/heyawake/seed0/model.safetensors b/trm/heyawake/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f1b3e2e55ccea292242b1b0457d79019a8864358 --- /dev/null +++ b/trm/heyawake/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bd9f92e94524ca8ff98738328a5daa64bdbb0721792242b2c1a5cf4d882f2f7 +size 28886576 diff --git a/trm/heyawake/seed1/config.json b/trm/heyawake/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d4a8c72610623ed37707d16b0d617a3a3b7ebd21 --- /dev/null +++ b/trm/heyawake/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/heyawake/seed1/model.safetensors b/trm/heyawake/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cb3b5d61ffc0a7b28b53ec12d54fb793346ab456 --- /dev/null +++ b/trm/heyawake/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90a424529419903746e2be083da276d926803c9e201f00fa8b600fb745ee407a +size 28886576 diff --git a/trm/heyawake/seed2/config.json b/trm/heyawake/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d4a8c72610623ed37707d16b0d617a3a3b7ebd21 --- /dev/null +++ b/trm/heyawake/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 394, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/heyawake/seed2/model.safetensors b/trm/heyawake/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b747c8896b6fc2c38497030daeb0242aabd78dbe --- /dev/null +++ b/trm/heyawake/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ce4c94dc6d08a136eb447ad93d06f2e7a6e07bd4f4067592e2630b15ced42d5 +size 28886576 diff --git a/trm/lightup/seed0/config.json b/trm/lightup/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..689b5adc2a6664702cdab40d33707d7ef7797f2a --- /dev/null +++ b/trm/lightup/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/lightup/seed0/model.safetensors b/trm/lightup/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5f56132f2aa0a33ea15e00109889bca0a7b38ecf --- /dev/null +++ b/trm/lightup/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c34534a2a21ba8dbec423f87ef69f97fa83e11974020d8159d14f87e8abfb47d +size 27309616 diff --git a/trm/lightup/seed1/config.json b/trm/lightup/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..689b5adc2a6664702cdab40d33707d7ef7797f2a --- /dev/null +++ b/trm/lightup/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/lightup/seed1/model.safetensors b/trm/lightup/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e02f1766769c54a506ee6d218ea1f5477d3a073a --- /dev/null +++ b/trm/lightup/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f2e147d91ab734d3500322568bb0d2923d81292c0cdf5da93fad693ff27f007 +size 27309616 diff --git a/trm/lightup/seed2/config.json b/trm/lightup/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..689b5adc2a6664702cdab40d33707d7ef7797f2a --- /dev/null +++ b/trm/lightup/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 9, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/lightup/seed2/model.safetensors b/trm/lightup/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..44f97b70eca13cf60eaa267d777319ee9af5c0c8 --- /dev/null +++ b/trm/lightup/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1a7abb6af31aeaf76ceafb8260642abd0bba637d38b7fea5a5f4f01cdfc9340 +size 27309616 diff --git a/trm/maze/seed0/config.json b/trm/maze/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dbb81377b84d75952b742159ba2c61831e81ef3a --- /dev/null +++ b/trm/maze/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/maze/seed0/model.safetensors b/trm/maze/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..02518b8a0b9fc305c185ea4f94688a3b1b1badcb --- /dev/null +++ b/trm/maze/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17a66c2fa138020045dcd4123771ad98b0b97f6ebd159162727528e8562e1a93 +size 27297328 diff --git a/trm/maze/seed1/config.json b/trm/maze/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dbb81377b84d75952b742159ba2c61831e81ef3a --- /dev/null +++ b/trm/maze/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/maze/seed1/model.safetensors b/trm/maze/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4fefecf57f13cbb5231e0727529a2b68387cc25c --- /dev/null +++ b/trm/maze/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b58def59db32b6708fa6749e5884da49208cc3c4e68ddaf5b1a2094887d48854 +size 27297328 diff --git a/trm/maze/seed2/config.json b/trm/maze/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dbb81377b84d75952b742159ba2c61831e81ef3a --- /dev/null +++ b/trm/maze/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/maze/seed2/model.safetensors b/trm/maze/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a528cc79761f854d2259da3dfb71e84321a86860 --- /dev/null +++ b/trm/maze/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:874dbcc7be2f9391308a1b28787e51964d6a2bd79a0006927d4ab446b792691a +size 27297328 diff --git a/trm/nurikabe/seed0/config.json b/trm/nurikabe/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cc282e4c1d7d2d15b81d7213f4aa08e03c1f1ada --- /dev/null +++ b/trm/nurikabe/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/nurikabe/seed0/model.safetensors b/trm/nurikabe/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..947710dc35770d021902f9d24e6f66e615eb5874 --- /dev/null +++ b/trm/nurikabe/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:439e2af53c8df57a1da2cc4bb2c72427493fd05feb2b909b27aab247af5b807e +size 27502128 diff --git a/trm/nurikabe/seed1/config.json b/trm/nurikabe/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cc282e4c1d7d2d15b81d7213f4aa08e03c1f1ada --- /dev/null +++ b/trm/nurikabe/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/nurikabe/seed1/model.safetensors b/trm/nurikabe/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e5d3186971b95977e4087ca1ae757b6275e66b3e --- /dev/null +++ b/trm/nurikabe/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a46cbf9317c7388775d0a6ea74fdd9e3de372794dc7406557e1b56dfa5fafb51 +size 27502128 diff --git a/trm/nurikabe/seed2/config.json b/trm/nurikabe/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cc282e4c1d7d2d15b81d7213f4aa08e03c1f1ada --- /dev/null +++ b/trm/nurikabe/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 56, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/nurikabe/seed2/model.safetensors b/trm/nurikabe/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..268eed7bab62532d3fce5413c9387be255dbc386 --- /dev/null +++ b/trm/nurikabe/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecaa9ceb1237e16ad94e5986e7cda2584ec3b2f6faf7637fa32226269fe401ed +size 27502128 diff --git a/trm/sudoku/seed0/config.json b/trm/sudoku/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..12141c5dc222a0dfe86374961fcc5945498a0713 --- /dev/null +++ b/trm/sudoku/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "none", + "mlp_t": true, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/sudoku/seed0/model.safetensors b/trm/sudoku/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d682db2f8f9c6e4b2e6b4ff330731cc4af2abbd8 --- /dev/null +++ b/trm/sudoku/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a967559a191a5071dc0a6d9db7e896c18d69948fc97e43f9594c8717c49f8b4f +size 20121128 diff --git a/trm/sudoku/seed1/config.json b/trm/sudoku/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..12141c5dc222a0dfe86374961fcc5945498a0713 --- /dev/null +++ b/trm/sudoku/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "none", + "mlp_t": true, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/sudoku/seed1/model.safetensors b/trm/sudoku/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..36f98d878afdf9b0907fe0caccea9158b318fc49 --- /dev/null +++ b/trm/sudoku/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c01e95ee465c9aa74c6cb352005549ed12cd09cb886f6ea36c2f1165fac024d +size 20121128 diff --git a/trm/sudoku/seed2/config.json b/trm/sudoku/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..12141c5dc222a0dfe86374961fcc5945498a0713 --- /dev/null +++ b/trm/sudoku/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 81, + "vocab_size": 11, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "none", + "mlp_t": true, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/sudoku/seed2/model.safetensors b/trm/sudoku/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6d110f778ffce522bf5ced67c0926bc270a334e1 --- /dev/null +++ b/trm/sudoku/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da83ee9bc4dddc990655890d6f8d5e6ef97916af75be85ffca9aa960ed481584 +size 20121128 diff --git a/trm/tapa/seed0/config.json b/trm/tapa/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebee48c818674755d75a83fbd7f69818976f934c --- /dev/null +++ b/trm/tapa/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/tapa/seed0/model.safetensors b/trm/tapa/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..86b8a52503080bc0f931dd9c795c8330915145ce --- /dev/null +++ b/trm/tapa/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:264666c616f887cd1f6ced6a8f6d2fc9e966a42e25a33ed941fb22576515d65c +size 27440688 diff --git a/trm/tapa/seed1/config.json b/trm/tapa/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebee48c818674755d75a83fbd7f69818976f934c --- /dev/null +++ b/trm/tapa/seed1/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/tapa/seed1/model.safetensors b/trm/tapa/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2755c503d008ec251717ba8ac61a0c4a6808c51d --- /dev/null +++ b/trm/tapa/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e37437ecf3a6c04140e08e9f35cd619862169a143c65847b4be3d41f6a585bf5 +size 27440688 diff --git a/trm/tapa/seed2/config.json b/trm/tapa/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebee48c818674755d75a83fbd7f69818976f934c --- /dev/null +++ b/trm/tapa/seed2/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 676, + "vocab_size": 41, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 6, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/trm/tapa/seed2/model.safetensors b/trm/tapa/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c2d9581229ec17081f10fe67e3cf21b92d0eaca1 --- /dev/null +++ b/trm/tapa/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8978da78cb3ac8f4f0cb624428225c751ee260ac64005f483c7aa5d98e87027 +size 27440688 diff --git a/verifier/maze/seed0/config.json b/verifier/maze/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dbb81377b84d75952b742159ba2c61831e81ef3a --- /dev/null +++ b/verifier/maze/seed0/config.json @@ -0,0 +1,24 @@ +{ + "model": { + "seq_len": 900, + "vocab_size": 6, + "num_puzzle_identifiers": 1, + "puzzle_emb_ndim": 512, + "puzzle_emb_len": 16, + "H_cycles": 3, + "L_cycles": 4, + "L_layers": 2, + "hidden_size": 512, + "expansion": 4.0, + "num_heads": 8, + "pos_encodings": "rope", + "mlp_t": false, + "rms_norm_eps": 1e-05, + "rope_theta": 10000.0, + "halt_max_steps": 16, + "halt_exploration_prob": 0.1, + "forward_dtype": "bfloat16", + "batch_size": 1 + }, + "provenance": {} +} \ No newline at end of file diff --git a/verifier/maze/seed0/model.safetensors b/verifier/maze/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..badd690e236c834b4e1d51e16cd831427a70f224 --- /dev/null +++ b/verifier/maze/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e93f27f8380f794eb23bb5362b85b42147b1247726985b112363f2ccf072922 +size 27297328