{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "SUCCESS", "10": "FAIL", "100": "NOT_ALIVE", "101": "SUCCESS", "102": "NOT_ALIVE", "103": "NON_PARTICIPATING", "104": "NON_PARTICIPATING", "105": "NOT_ALIVE", "106": "SUCCESS", "107": "SUCCESS", "108": "SUCCESS", "109": "NOT_ALIVE", "11": "SUCCESS", "110": "FAIL", "111": "NOT_ALIVE", "112": "SUCCESS", "113": "NOT_ALIVE", "114": "NOT_ALIVE", "115": "NON_PARTICIPATING", "116": "NON_PARTICIPATING", "117": "SUCCESS", "118": "NOT_ALIVE", "119": "SUCCESS", "12": "NOT_ALIVE", "120": "NON_PARTICIPATING", "121": "SUCCESS", "122": "SUCCESS", "123": "SUCCESS", "124": "NOT_ALIVE", "125": "NOT_ALIVE", "126": "SUCCESS", "127": "NON_PARTICIPATING", "128": "SUCCESS", "129": "FAIL", "13": "NOT_ALIVE", "130": "SUCCESS", "131": "NON_PARTICIPATING", "132": "SUCCESS", "133": "SUCCESS", "134": "NOT_ALIVE", "135": "SUCCESS", "136": "SUCCESS", "137": "SUCCESS", "138": "NON_PARTICIPATING", "139": "SUCCESS", "14": "NOT_ALIVE", "140": "NON_PARTICIPATING", "141": "NOT_ALIVE", "142": "FAIL", "143": "FAIL", "144": "NON_PARTICIPATING", "145": "SUCCESS", "146": "NON_PARTICIPATING", "147": "FAIL", "148": "NON_PARTICIPATING", "149": "NOT_ALIVE", "15": "SUCCESS", "150": "NOT_ALIVE", "151": "NON_PARTICIPATING", "152": "NOT_ALIVE", "153": "SUCCESS", "154": "NON_PARTICIPATING", "155": "NON_PARTICIPATING", "156": "NOT_ALIVE", "157": "NON_PARTICIPATING", "158": "NON_PARTICIPATING", "159": "SUCCESS", "16": "NOT_ALIVE", "160": "NON_PARTICIPATING", "161": "SUCCESS", "162": "NON_PARTICIPATING", "163": "SUCCESS", "164": "NON_PARTICIPATING", "165": "SUCCESS", "166": "NOT_ALIVE", "167": "NON_PARTICIPATING", "168": "SUCCESS", "169": "SUCCESS", "17": "NOT_ALIVE", "170": "SUCCESS", "171": "SUCCESS", "172": "SUCCESS", "173": "SUCCESS", "174": "SUCCESS", "175": "SUCCESS", "176": "NOT_ALIVE", "177": "SUCCESS", "178": "NON_PARTICIPATING", "179": "SUCCESS", "18": "NOT_ALIVE", "180": "SUCCESS", "181": "SUCCESS", "182": "SUCCESS", "183": "FAIL", "184": "SUCCESS", "185": "FAIL", "186": "NON_PARTICIPATING", "187": "NOT_ALIVE", "188": "SUCCESS", "189": "FAIL", "19": "FAIL", "190": "NON_PARTICIPATING", "191": "FAIL", "192": "SUCCESS", "193": "FAIL", "194": "NON_PARTICIPATING", "195": "SUCCESS", "196": "NON_PARTICIPATING", "197": "NON_PARTICIPATING", "198": "NON_PARTICIPATING", "199": "SUCCESS", "2": "NOT_ALIVE", "20": "NOT_ALIVE", "200": "SUCCESS", "201": "NOT_ALIVE", "202": "NON_PARTICIPATING", "203": "SUCCESS", "204": "SUCCESS", "205": "SUCCESS", "206": "NON_PARTICIPATING", "207": "NON_PARTICIPATING", "208": "NON_PARTICIPATING", "209": "SUCCESS", "21": "SUCCESS", "210": "NON_PARTICIPATING", "211": "SUCCESS", "212": "SUCCESS", "213": "NON_PARTICIPATING", "214": "SUCCESS", "215": "SUCCESS", "216": "SUCCESS", "217": "NON_PARTICIPATING", "218": "NON_PARTICIPATING", "219": "SUCCESS", "22": "FAIL", "220": "SUCCESS", "221": "SUCCESS", "222": "SUCCESS", "223": "SUCCESS", "224": "NON_PARTICIPATING", "225": "SUCCESS", "226": "NOT_ALIVE", "227": "NOT_ALIVE", "228": "SUCCESS", "229": "NON_PARTICIPATING", "23": "NOT_ALIVE", "230": "SUCCESS", "231": "NOT_ALIVE", "232": "NOT_ALIVE", "233": "SUCCESS", "234": "SUCCESS", "235": "SUCCESS", "236": "SUCCESS", "237": "NON_PARTICIPATING", "238": "SUCCESS", "239": "SUCCESS", "24": "NOT_ALIVE", "240": "NON_PARTICIPATING", "241": "NON_PARTICIPATING", "242": "SUCCESS", "243": "NOT_ALIVE", "244": "SUCCESS", "245": "SUCCESS", "246": "NON_PARTICIPATING", "247": "FAIL", "248": "FAIL", "249": "NOT_ALIVE", "25": "SUCCESS", "250": "FAIL", "251": "NOT_ALIVE", "252": "SUCCESS", "253": "SUCCESS", "254": "NOT_ALIVE", "255": "NON_PARTICIPATING", "26": "SUCCESS", "27": "FAIL", "28": "SUCCESS", "29": "SUCCESS", "3": "SUCCESS", "30": "FAIL", "31": "SUCCESS", "32": "NOT_ALIVE", "33": "NOT_ALIVE", "34": "SUCCESS", "35": "FAIL", "36": "SUCCESS", "37": "NOT_ALIVE", "38": "NOT_ALIVE", "39": "NON_PARTICIPATING", "4": "NOT_ALIVE", "40": "NON_PARTICIPATING", "41": "NON_PARTICIPATING", "42": "NOT_ALIVE", "43": "NON_PARTICIPATING", "44": "SUCCESS", "45": "NOT_ALIVE", "46": "FAIL", "47": "SUCCESS", "48": "NOT_ALIVE", "49": "NOT_ALIVE", "5": "SUCCESS", "50": "FAIL", "51": "SUCCESS", "52": "SUCCESS", "53": "SUCCESS", "54": "SUCCESS", "55": "NON_PARTICIPATING", "56": "NON_PARTICIPATING", "57": "SUCCESS", "58": "SUCCESS", "59": "NON_PARTICIPATING", "6": "NOT_ALIVE", "60": "SUCCESS", "61": "NON_PARTICIPATING", "62": "NOT_ALIVE", "63": "SUCCESS", "64": "SUCCESS", "65": "NOT_ALIVE", "66": "NOT_ALIVE", "67": "NOT_ALIVE", "68": "FAIL", "69": "NON_PARTICIPATING", "7": "NOT_ALIVE", "70": "NOT_ALIVE", "71": "NON_PARTICIPATING", "72": "NON_PARTICIPATING", "73": "NON_PARTICIPATING", "74": "SUCCESS", "75": "NON_PARTICIPATING", "76": "NON_PARTICIPATING", "77": "NON_PARTICIPATING", "78": "SUCCESS", "79": "SUCCESS", "8": "NOT_ALIVE", "80": "NON_PARTICIPATING", "81": "NOT_ALIVE", "82": "SUCCESS", "83": "SUCCESS", "84": "NOT_ALIVE", "85": "NON_PARTICIPATING", "86": "NOT_ALIVE", "87": "FAIL", "88": "NON_PARTICIPATING", "89": "NON_PARTICIPATING", "9": "NOT_ALIVE", "90": "SUCCESS", "91": "SUCCESS", "92": "NOT_ALIVE", "93": "NON_PARTICIPATING", "94": "NOT_ALIVE", "95": "FAIL", "96": "SUCCESS", "97": "SUCCESS", "98": "NON_PARTICIPATING", "99": "NON_PARTICIPATING" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6380445, 6380462 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 26, "intermediate_size": 5632, "last_allreduce_block": 6380215, "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 }