{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "NON_PARTICIPATING", "10": "NOT_ALIVE", "100": "NOT_ALIVE", "101": "NOT_ALIVE", "102": "NOT_ALIVE", "103": "SUCCESS", "104": "NOT_ALIVE", "105": "NON_PARTICIPATING", "106": "NOT_ALIVE", "107": "NOT_ALIVE", "108": "NON_PARTICIPATING", "109": "NON_PARTICIPATING", "11": "NOT_ALIVE", "110": "SUCCESS", "111": "SUCCESS", "112": "NON_PARTICIPATING", "113": "NON_PARTICIPATING", "114": "NON_PARTICIPATING", "115": "NOT_ALIVE", "116": "NON_PARTICIPATING", "117": "NOT_ALIVE", "118": "NOT_ALIVE", "119": "NOT_ALIVE", "12": "NON_PARTICIPATING", "120": "NOT_ALIVE", "121": "NOT_ALIVE", "122": "SUCCESS", "123": "SUCCESS", "124": "SUCCESS", "125": "NON_PARTICIPATING", "126": "NON_PARTICIPATING", "127": "NOT_ALIVE", "128": "NOT_ALIVE", "129": "NON_PARTICIPATING", "13": "NON_PARTICIPATING", "130": "SUCCESS", "131": "NOT_ALIVE", "132": "NOT_ALIVE", "133": "SUCCESS", "134": "NON_PARTICIPATING", "135": "NON_PARTICIPATING", "136": "SUCCESS", "137": "NOT_ALIVE", "138": "NOT_ALIVE", "139": "NON_PARTICIPATING", "14": "NON_PARTICIPATING", "140": "NOT_ALIVE", "141": "SUCCESS", "142": "SUCCESS", "143": "NON_PARTICIPATING", "144": "NON_PARTICIPATING", "145": "NON_PARTICIPATING", "146": "SUCCESS", "147": "NOT_ALIVE", "148": "SUCCESS", "149": "NON_PARTICIPATING", "15": "NON_PARTICIPATING", "150": "SUCCESS", "151": "NOT_ALIVE", "152": "NON_PARTICIPATING", "153": "SUCCESS", "154": "NON_PARTICIPATING", "155": "SUCCESS", "156": "NOT_ALIVE", "157": "NON_PARTICIPATING", "158": "NOT_ALIVE", "159": "NOT_ALIVE", "16": "NON_PARTICIPATING", "160": "NON_PARTICIPATING", "161": "NOT_ALIVE", "162": "NON_PARTICIPATING", "163": "NOT_ALIVE", "164": "SUCCESS", "165": "SUCCESS", "166": "NOT_ALIVE", "167": "NON_PARTICIPATING", "168": "SUCCESS", "169": "NON_PARTICIPATING", "17": "NON_PARTICIPATING", "170": "NON_PARTICIPATING", "171": "NOT_ALIVE", "172": "NOT_ALIVE", "173": "NOT_ALIVE", "174": "NOT_ALIVE", "175": "NOT_ALIVE", "176": "NOT_ALIVE", "177": "NON_PARTICIPATING", "178": "NOT_ALIVE", "179": "NOT_ALIVE", "18": "SUCCESS", "180": "NON_PARTICIPATING", "181": "SUCCESS", "182": "NON_PARTICIPATING", "183": "SUCCESS", "184": "NOT_ALIVE", "185": "NON_PARTICIPATING", "186": "NOT_ALIVE", "187": "SUCCESS", "188": "NOT_ALIVE", "189": "NOT_ALIVE", "19": "NON_PARTICIPATING", "190": "NON_PARTICIPATING", "191": "NOT_ALIVE", "192": "NON_PARTICIPATING", "193": "NON_PARTICIPATING", "194": "NOT_ALIVE", "195": "NOT_ALIVE", "196": "NOT_ALIVE", "197": "NON_PARTICIPATING", "198": "NOT_ALIVE", "199": "NON_PARTICIPATING", "2": "NOT_ALIVE", "20": "NOT_ALIVE", "200": "NON_PARTICIPATING", "201": "NON_PARTICIPATING", "202": "NON_PARTICIPATING", "203": "NON_PARTICIPATING", "204": "NON_PARTICIPATING", "205": "NOT_ALIVE", "206": "SUCCESS", "207": "NOT_ALIVE", "208": "NOT_ALIVE", "209": "SUCCESS", "21": "SUCCESS", "210": "SUCCESS", "211": "NOT_ALIVE", "212": "NOT_ALIVE", "213": "SUCCESS", "214": "NON_PARTICIPATING", "215": "SUCCESS", "216": "NOT_ALIVE", "217": "NOT_ALIVE", "218": "NON_PARTICIPATING", "219": "NON_PARTICIPATING", "22": "NON_PARTICIPATING", "220": "NON_PARTICIPATING", "221": "SUCCESS", "222": "NON_PARTICIPATING", "223": "NOT_ALIVE", "224": "NOT_ALIVE", "225": "SUCCESS", "226": "NOT_ALIVE", "227": "SUCCESS", "228": "NON_PARTICIPATING", "229": "NOT_ALIVE", "23": "NON_PARTICIPATING", "230": "NOT_ALIVE", "231": "NON_PARTICIPATING", "232": "NON_PARTICIPATING", "233": "SUCCESS", "234": "NOT_ALIVE", "235": "NOT_ALIVE", "236": "NOT_ALIVE", "237": "NOT_ALIVE", "238": "SUCCESS", "239": "NON_PARTICIPATING", "24": "SUCCESS", "240": "NOT_ALIVE", "241": "NON_PARTICIPATING", "242": "SUCCESS", "243": "NON_PARTICIPATING", "244": "NOT_ALIVE", "245": "NOT_ALIVE", "246": "NON_PARTICIPATING", "247": "NON_PARTICIPATING", "248": "NON_PARTICIPATING", "249": "NOT_ALIVE", "25": "SUCCESS", "250": "NON_PARTICIPATING", "251": "NOT_ALIVE", "252": "NON_PARTICIPATING", "253": "NOT_ALIVE", "254": "NON_PARTICIPATING", "255": "NON_PARTICIPATING", "26": "NON_PARTICIPATING", "27": "SUCCESS", "28": "SUCCESS", "29": "NOT_ALIVE", "3": "NON_PARTICIPATING", "30": "SUCCESS", "31": "NON_PARTICIPATING", "32": "SUCCESS", "33": "NOT_ALIVE", "34": "SUCCESS", "35": "SUCCESS", "36": "NOT_ALIVE", "37": "NON_PARTICIPATING", "38": "SUCCESS", "39": "SUCCESS", "4": "NOT_ALIVE", "40": "SUCCESS", "41": "SUCCESS", "42": "SUCCESS", "43": "NOT_ALIVE", "44": "NON_PARTICIPATING", "45": "NON_PARTICIPATING", "46": "NON_PARTICIPATING", "47": "NON_PARTICIPATING", "48": "SUCCESS", "49": "NON_PARTICIPATING", "5": "NOT_ALIVE", "50": "NOT_ALIVE", "51": "NON_PARTICIPATING", "52": "NOT_ALIVE", "53": "NOT_ALIVE", "54": "SUCCESS", "55": "NON_PARTICIPATING", "56": "NOT_ALIVE", "57": "SUCCESS", "58": "SUCCESS", "59": "NOT_ALIVE", "6": "NOT_ALIVE", "60": "NON_PARTICIPATING", "61": "NOT_ALIVE", "62": "NOT_ALIVE", "63": "NON_PARTICIPATING", "64": "NON_PARTICIPATING", "65": "NOT_ALIVE", "66": "NON_PARTICIPATING", "67": "NON_PARTICIPATING", "68": "NOT_ALIVE", "69": "NON_PARTICIPATING", "7": "SUCCESS", "70": "NOT_ALIVE", "71": "NON_PARTICIPATING", "72": "NOT_ALIVE", "73": "SUCCESS", "74": "SUCCESS", "75": "SUCCESS", "76": "NON_PARTICIPATING", "77": "NOT_ALIVE", "78": "NON_PARTICIPATING", "79": "NON_PARTICIPATING", "8": "NON_PARTICIPATING", "80": "NON_PARTICIPATING", "81": "NOT_ALIVE", "82": "NOT_ALIVE", "83": "NOT_ALIVE", "84": "NOT_ALIVE", "85": "NOT_ALIVE", "86": "SUCCESS", "87": "NON_PARTICIPATING", "88": "NOT_ALIVE", "89": "NON_PARTICIPATING", "9": "NOT_ALIVE", "90": "SUCCESS", "91": "NOT_ALIVE", "92": "SUCCESS", "93": "NON_PARTICIPATING", "94": "NON_PARTICIPATING", "95": "NOT_ALIVE", "96": "NON_PARTICIPATING", "97": "SUCCESS", "98": "SUCCESS", "99": "NOT_ALIVE" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 5907297, 5907329 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 15, "intermediate_size": 5632, "last_allreduce_block": 5906842, "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 }