{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "NOT_ALIVE", "10": "NOT_ALIVE", "100": "NON_PARTICIPATING", "101": "NOT_ALIVE", "102": "NON_PARTICIPATING", "103": "NOT_ALIVE", "104": "NON_PARTICIPATING", "105": "NOT_ALIVE", "106": "NOT_ALIVE", "107": "NON_PARTICIPATING", "108": "NOT_ALIVE", "109": "NON_PARTICIPATING", "11": "NOT_ALIVE", "110": "NON_PARTICIPATING", "111": "NOT_ALIVE", "112": "SUCCESS", "113": "NOT_ALIVE", "114": "NOT_ALIVE", "115": "NOT_ALIVE", "116": "NOT_ALIVE", "117": "NOT_ALIVE", "118": "NOT_ALIVE", "119": "NOT_ALIVE", "12": "NOT_ALIVE", "120": "NOT_ALIVE", "121": "SUCCESS", "122": "NON_PARTICIPATING", "123": "SUCCESS", "124": "NOT_ALIVE", "125": "NOT_ALIVE", "126": "FAIL", "127": "NON_PARTICIPATING", "128": "SUCCESS", "129": "NON_PARTICIPATING", "13": "NOT_ALIVE", "130": "NON_PARTICIPATING", "131": "NOT_ALIVE", "132": "SUCCESS", "133": "SUCCESS", "134": "NON_PARTICIPATING", "135": "SUCCESS", "136": "SUCCESS", "137": "FAIL", "138": "NON_PARTICIPATING", "139": "NOT_ALIVE", "14": "NOT_ALIVE", "140": "NON_PARTICIPATING", "141": "NOT_ALIVE", "142": "NON_PARTICIPATING", "143": "NOT_ALIVE", "144": "NON_PARTICIPATING", "145": "SUCCESS", "146": "NON_PARTICIPATING", "147": "NOT_ALIVE", "148": "NON_PARTICIPATING", "149": "SUCCESS", "15": "SUCCESS", "150": "SUCCESS", "151": "NON_PARTICIPATING", "152": "FAIL", "153": "SUCCESS", "154": "NOT_ALIVE", "155": "FAIL", "156": "NOT_ALIVE", "157": "NON_PARTICIPATING", "158": "NON_PARTICIPATING", "159": "NOT_ALIVE", "16": "NOT_ALIVE", "160": "NON_PARTICIPATING", "161": "NON_PARTICIPATING", "162": "NON_PARTICIPATING", "163": "NOT_ALIVE", "164": "FAIL", "165": "SUCCESS", "166": "NOT_ALIVE", "167": "NOT_ALIVE", "168": "NOT_ALIVE", "169": "SUCCESS", "17": "NOT_ALIVE", "170": "SUCCESS", "171": "NON_PARTICIPATING", "172": "NON_PARTICIPATING", "173": "NOT_ALIVE", "174": "SUCCESS", "175": "NON_PARTICIPATING", "176": "NOT_ALIVE", "177": "SUCCESS", "178": "NON_PARTICIPATING", "179": "SUCCESS", "18": "NOT_ALIVE", "180": "NON_PARTICIPATING", "181": "FAIL", "182": "NOT_ALIVE", "183": "NOT_ALIVE", "184": "NOT_ALIVE", "185": "NON_PARTICIPATING", "186": "FAIL", "187": "NOT_ALIVE", "188": "FAIL", "189": "SUCCESS", "19": "NOT_ALIVE", "190": "NON_PARTICIPATING", "191": "SUCCESS", "192": "NON_PARTICIPATING", "193": "NOT_ALIVE", "194": "NON_PARTICIPATING", "195": "NON_PARTICIPATING", "196": "SUCCESS", "197": "NOT_ALIVE", "198": "NOT_ALIVE", "199": "SUCCESS", "2": "NOT_ALIVE", "20": "NOT_ALIVE", "200": "NOT_ALIVE", "201": "NOT_ALIVE", "202": "FAIL", "203": "NOT_ALIVE", "204": "NON_PARTICIPATING", "205": "NOT_ALIVE", "206": "NON_PARTICIPATING", "207": "NOT_ALIVE", "208": "NOT_ALIVE", "209": "NOT_ALIVE", "21": "NOT_ALIVE", "210": "NOT_ALIVE", "211": "SUCCESS", "212": "NOT_ALIVE", "213": "NON_PARTICIPATING", "214": "NON_PARTICIPATING", "215": "FAIL", "216": "SUCCESS", "217": "SUCCESS", "218": "FAIL", "219": "FAIL", "22": "NOT_ALIVE", "220": "FAIL", "221": "SUCCESS", "222": "NOT_ALIVE", "223": "NOT_ALIVE", "224": "NON_PARTICIPATING", "225": "SUCCESS", "226": "NOT_ALIVE", "227": "NON_PARTICIPATING", "228": "SUCCESS", "229": "NON_PARTICIPATING", "23": "NOT_ALIVE", "230": "NOT_ALIVE", "231": "NOT_ALIVE", "232": "NOT_ALIVE", "233": "NON_PARTICIPATING", "234": "SUCCESS", "235": "NOT_ALIVE", "236": "SUCCESS", "237": "NON_PARTICIPATING", "238": "SUCCESS", "239": "NON_PARTICIPATING", "24": "SUCCESS", "240": "NON_PARTICIPATING", "241": "NON_PARTICIPATING", "242": "NON_PARTICIPATING", "243": "NON_PARTICIPATING", "244": "NON_PARTICIPATING", "245": "NON_PARTICIPATING", "246": "NON_PARTICIPATING", "247": "NON_PARTICIPATING", "248": "NON_PARTICIPATING", "249": "NOT_ALIVE", "25": "SUCCESS", "250": "NON_PARTICIPATING", "251": "NON_PARTICIPATING", "252": "SUCCESS", "253": "SUCCESS", "254": "NOT_ALIVE", "255": "NON_PARTICIPATING", "26": "FAIL", "27": "SUCCESS", "28": "NON_PARTICIPATING", "29": "SUCCESS", "3": "FAIL", "30": "SUCCESS", "31": "NON_PARTICIPATING", "32": "NON_PARTICIPATING", "33": "NOT_ALIVE", "34": "NOT_ALIVE", "35": "NON_PARTICIPATING", "36": "FAIL", "37": "NOT_ALIVE", "38": "SUCCESS", "39": "NOT_ALIVE", "4": "NOT_ALIVE", "40": "NON_PARTICIPATING", "41": "FAIL", "42": "NOT_ALIVE", "43": "NON_PARTICIPATING", "44": "NON_PARTICIPATING", "45": "NOT_ALIVE", "46": "NON_PARTICIPATING", "47": "NOT_ALIVE", "48": "NON_PARTICIPATING", "49": "NON_PARTICIPATING", "5": "NOT_ALIVE", "50": "NOT_ALIVE", "51": "NON_PARTICIPATING", "52": "SUCCESS", "53": "SUCCESS", "54": "NON_PARTICIPATING", "55": "NON_PARTICIPATING", "56": "NOT_ALIVE", "57": "NOT_ALIVE", "58": "NOT_ALIVE", "59": "NOT_ALIVE", "6": "NOT_ALIVE", "60": "NOT_ALIVE", "61": "NON_PARTICIPATING", "62": "NOT_ALIVE", "63": "FAIL", "64": "NOT_ALIVE", "65": "NOT_ALIVE", "66": "NON_PARTICIPATING", "67": "NON_PARTICIPATING", "68": "NOT_ALIVE", "69": "NON_PARTICIPATING", "7": "NOT_ALIVE", "70": "NOT_ALIVE", "71": "NOT_ALIVE", "72": "NOT_ALIVE", "73": "NOT_ALIVE", "74": "SUCCESS", "75": "NOT_ALIVE", "76": "NOT_ALIVE", "77": "NON_PARTICIPATING", "78": "NON_PARTICIPATING", "79": "SUCCESS", "8": "NOT_ALIVE", "80": "NON_PARTICIPATING", "81": "NOT_ALIVE", "82": "NON_PARTICIPATING", "83": "NOT_ALIVE", "84": "NON_PARTICIPATING", "85": "NOT_ALIVE", "86": "NOT_ALIVE", "87": "FAIL", "88": "NON_PARTICIPATING", "89": "NON_PARTICIPATING", "9": "NOT_ALIVE", "90": "NON_PARTICIPATING", "91": "NON_PARTICIPATING", "92": "NOT_ALIVE", "93": "NOT_ALIVE", "94": "NOT_ALIVE", "95": "NOT_ALIVE", "96": "NOT_ALIVE", "97": "FAIL", "98": "NON_PARTICIPATING", "99": "NON_PARTICIPATING" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6372877, 6372895, 6372911, 6372929, 6372949 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 19, "intermediate_size": 5632, "last_allreduce_block": 6372758, "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 }