share gpt2 checkpoint

config.json

@@ -0,0 +1 @@
+{"model": {"gpt2": {"name": "gpt2", "gpt2_type": "gpt2", "from_pretrained": true, "lm_head": true}, "name": "bc_lm", "dataset": {"name": "hackernews_list_dataset", "cache_id": "d_train"}, "transition_weight": 0.0, "load": {"name": "bc_lm", "checkpoint_path": null, "strict_load": true}}, "train_dataset": {"data": {"name": "hackernews_rl_dataset", "path": "data/hackernews_rl_dataset/", "cache_path": null, "reward_shift": 0.0, "reward_scale": 1.0, "reward_f": {"name": "score_human_reward", "job_descriptions_path": "data/hackernews_rl_dataset/", "index_path": "data/hackernews_rl_dataset/train_idxs.json"}, "index_path": "data/hackernews_rl_dataset/train_idxs.json", "cache_id": "train_raw_data"}, "token_reward": {"name": "constant_token_reward", "c": 0.0}, "name": "hackernews_list_dataset", "max_len": 1024, "cuttoff": null, "resample_timeout": 0.0, "cache_id": "d_train", "include_parent": true}, "eval_dataset": {"data": {"name": "hackernews_rl_dataset", "path": "data/hackernews_rl_dataset/", "cache_path": null, "reward_shift": 0.0, "reward_scale": 1.0, "reward_f": {"name": "score_human_reward", "job_descriptions_path": "data/hackernews_rl_dataset/", "index_path": "data/hackernews_rl_dataset/test_idxs.json"}, "index_path": "data/hackernews_rl_dataset/test_idxs.json", "cache_id": "test_raw_data"}, "token_reward": {"name": "constant_token_reward", "c": 0.0}, "name": "hackernews_list_dataset", "max_len": 256, "cuttoff": null, "resample_timeout": 0.0, "cache_id": "d_test", "include_parent": true}, "evaluator": {"env": {"name": "hackernews_env", "reward_shift": 0.0, "reward_scale": 1.0, "reward_f": {"name": "hackernews_reward"}, "data": {"name": "hackernews_rl_dataset", "cache_id": "test_raw_data"}, "include_parent": true}, "name": "bc_evaluator", "verbose": true, "kind": "sample", "generation_kwargs": {"max_generation_len": 256, "num_generations": 1}}, "train": {"save_checkpoint_dir": "/dccstor/autofair/bias_llm/Bias-ILQL/src/utils/../../outputs/hackernews/conditional_hackernews_official_bc_test_fix_q/", "optim_state_path": null, "epochs": 40, "dataloader_workers": 0, "bsize": 1, "grad_accum_steps": 64, "log_every": 256, "eval_every": 4096, "save_every": 8192, "max_checkpoints": 1, "eval_bsize": 1, "eval_batches": 16, "lr": 0.001, "weight_decay": 0.0, "max_steps": null, "loss": {}}, "wandb": {"use_wandb": false, "wandb_project": "hackernews_iql"}, "system": {"device": "cuda", "num_processes": 1, "use_fp16": false}}

model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0f7d486cadcd3ebbfed00971226fee183014987efefa34c1a6fe931caf6c41f
+size 510414733