{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "FAIL", "10": "NON_PARTICIPATING", "100": "NOT_ALIVE", "101": "NOT_ALIVE", "102": "SUCCESS", "103": "NON_PARTICIPATING", "104": "NOT_ALIVE", "105": "FAIL", "106": "NOT_ALIVE", "107": "NOT_ALIVE", "108": "NON_PARTICIPATING", "109": "FAIL", "11": "FAIL", "110": "FAIL", "111": "NOT_ALIVE", "112": "NON_PARTICIPATING", "113": "NON_PARTICIPATING", "114": "FAIL", "115": "NON_PARTICIPATING", "116": "FAIL", "117": "NOT_ALIVE", "118": "NOT_ALIVE", "119": "NON_PARTICIPATING", "12": "NOT_ALIVE", "120": "NOT_ALIVE", "121": "FAIL", "122": "FAIL", "123": "FAIL", "124": "NON_PARTICIPATING", "125": "FAIL", "126": "FAIL", "127": "NON_PARTICIPATING", "128": "FAIL", "129": "FAIL", "13": "NON_PARTICIPATING", "130": "NOT_ALIVE", "131": "NOT_ALIVE", "132": "NOT_ALIVE", "133": "NOT_ALIVE", "134": "NOT_ALIVE", "135": "NOT_ALIVE", "136": "NON_PARTICIPATING", "137": "NOT_ALIVE", "138": "FAIL", "139": "NON_PARTICIPATING", "14": "NOT_ALIVE", "140": "NON_PARTICIPATING", "141": "NOT_ALIVE", "142": "NOT_ALIVE", "143": "NON_PARTICIPATING", "144": "NON_PARTICIPATING", "145": "NOT_ALIVE", "146": "FAIL", "147": "NON_PARTICIPATING", "148": "NON_PARTICIPATING", "149": "NON_PARTICIPATING", "15": "NOT_ALIVE", "150": "NOT_ALIVE", "151": "NOT_ALIVE", "152": "NOT_ALIVE", "153": "FAIL", "154": "FAIL", "155": "NON_PARTICIPATING", "156": "FAIL", "157": "NON_PARTICIPATING", "158": "NON_PARTICIPATING", "159": "FAIL", "16": "NOT_ALIVE", "160": "NOT_ALIVE", "161": "FAIL", "162": "NOT_ALIVE", "163": "NON_PARTICIPATING", "164": "NOT_ALIVE", "165": "NOT_ALIVE", "166": "FAIL", "167": "NON_PARTICIPATING", "168": "NON_PARTICIPATING", "169": "FAIL", "17": "NOT_ALIVE", "170": "NOT_ALIVE", "171": "FAIL", "172": "NOT_ALIVE", "173": "NON_PARTICIPATING", "174": "NOT_ALIVE", "175": "SUCCESS", "176": "NOT_ALIVE", "177": "NON_PARTICIPATING", "178": "NOT_ALIVE", "179": "NOT_ALIVE", "18": "NOT_ALIVE", "180": "NON_PARTICIPATING", "181": "NOT_ALIVE", "182": "FAIL", "183": "NOT_ALIVE", "184": "FAIL", "185": "NOT_ALIVE", "186": "FAIL", "187": "NON_PARTICIPATING", "188": "NOT_ALIVE", "189": "NOT_ALIVE", "19": "NON_PARTICIPATING", "190": "NON_PARTICIPATING", "191": "NOT_ALIVE", "192": "NON_PARTICIPATING", "193": "FAIL", "194": "NOT_ALIVE", "195": "NOT_ALIVE", "196": "NON_PARTICIPATING", "197": "NON_PARTICIPATING", "198": "FAIL", "199": "NON_PARTICIPATING", "2": "NON_PARTICIPATING", "20": "FAIL", "200": "FAIL", "201": "NOT_ALIVE", "202": "NON_PARTICIPATING", "203": "NON_PARTICIPATING", "204": "NOT_ALIVE", "205": "NOT_ALIVE", "206": "FAIL", "207": "NOT_ALIVE", "208": "NON_PARTICIPATING", "209": "NON_PARTICIPATING", "21": "FAIL", "210": "FAIL", "211": "FAIL", "212": "FAIL", "213": "NON_PARTICIPATING", "214": "FAIL", "215": "NOT_ALIVE", "216": "NOT_ALIVE", "217": "NOT_ALIVE", "218": "NON_PARTICIPATING", "219": "FAIL", "22": "NOT_ALIVE", "220": "NON_PARTICIPATING", "221": "NON_PARTICIPATING", "222": "NON_PARTICIPATING", "223": "FAIL", "224": "NOT_ALIVE", "225": "NON_PARTICIPATING", "226": "NOT_ALIVE", "227": "NON_PARTICIPATING", "228": "NON_PARTICIPATING", "229": "NOT_ALIVE", "23": "NON_PARTICIPATING", "230": "NON_PARTICIPATING", "231": "NON_PARTICIPATING", "232": "FAIL", "233": "NON_PARTICIPATING", "234": "NON_PARTICIPATING", "235": "NOT_ALIVE", "236": "NON_PARTICIPATING", "237": "NON_PARTICIPATING", "238": "NON_PARTICIPATING", "239": "NON_PARTICIPATING", "24": "NON_PARTICIPATING", "240": "NOT_ALIVE", "241": "FAIL", "242": "NOT_ALIVE", "243": "NOT_ALIVE", "244": "NOT_ALIVE", "245": "SUCCESS", "246": "FAIL", "247": "NOT_ALIVE", "248": "NON_PARTICIPATING", "249": "NON_PARTICIPATING", "25": "SUCCESS", "250": "FAIL", "251": "NOT_ALIVE", "252": "FAIL", "253": "NOT_ALIVE", "254": "NON_PARTICIPATING", "255": "NON_PARTICIPATING", "26": "NOT_ALIVE", "27": "NOT_ALIVE", "28": "NON_PARTICIPATING", "29": "NON_PARTICIPATING", "3": "FAIL", "30": "NON_PARTICIPATING", "31": "FAIL", "32": "NOT_ALIVE", "33": "NOT_ALIVE", "34": "FAIL", "35": "SUCCESS", "36": "NON_PARTICIPATING", "37": "NON_PARTICIPATING", "38": "NON_PARTICIPATING", "39": "NON_PARTICIPATING", "4": "NON_PARTICIPATING", "40": "NON_PARTICIPATING", "41": "NOT_ALIVE", "42": "NON_PARTICIPATING", "43": "NOT_ALIVE", "44": "NON_PARTICIPATING", "45": "NON_PARTICIPATING", "46": "NOT_ALIVE", "47": "NON_PARTICIPATING", "48": "NOT_ALIVE", "49": "NOT_ALIVE", "5": "NON_PARTICIPATING", "50": "NON_PARTICIPATING", "51": "NOT_ALIVE", "52": "NON_PARTICIPATING", "53": "NON_PARTICIPATING", "54": "NOT_ALIVE", "55": "NOT_ALIVE", "56": "NON_PARTICIPATING", "57": "FAIL", "58": "NOT_ALIVE", "59": "NON_PARTICIPATING", "6": "NOT_ALIVE", "60": "NOT_ALIVE", "61": "FAIL", "62": "SUCCESS", "63": "NON_PARTICIPATING", "64": "FAIL", "65": "NON_PARTICIPATING", "66": "NON_PARTICIPATING", "67": "NON_PARTICIPATING", "68": "NOT_ALIVE", "69": "FAIL", "7": "NON_PARTICIPATING", "70": "NON_PARTICIPATING", "71": "NON_PARTICIPATING", "72": "NON_PARTICIPATING", "73": "NON_PARTICIPATING", "74": "FAIL", "75": "NON_PARTICIPATING", "76": "NOT_ALIVE", "77": "NOT_ALIVE", "78": "NOT_ALIVE", "79": "NOT_ALIVE", "8": "NON_PARTICIPATING", "80": "NON_PARTICIPATING", "81": "NOT_ALIVE", "82": "FAIL", "83": "NOT_ALIVE", "84": "FAIL", "85": "NON_PARTICIPATING", "86": "NOT_ALIVE", "87": "FAIL", "88": "NOT_ALIVE", "89": "NOT_ALIVE", "9": "NON_PARTICIPATING", "90": "NOT_ALIVE", "91": "NON_PARTICIPATING", "92": "SUCCESS", "93": "NOT_ALIVE", "94": "SUCCESS", "95": "NOT_ALIVE", "96": "SUCCESS", "97": "NOT_ALIVE", "98": "NON_PARTICIPATING", "99": "NON_PARTICIPATING" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6496814, 6496844 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 54, "intermediate_size": 5632, "last_allreduce_block": 6496409, "max_position_embeddings": 2048, "mlp_bias": false, "model_type": "llama", "num_attention_heads": 32, "num_hidden_layers": 22, "num_key_value_heads": 4, "pretraining_tp": 1, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 10000.0, "tie_word_embeddings": false, "torch_dtype": "float32", "transformers_version": "4.39.3", "use_cache": false, "vocab_size": 32000 }