{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "NON_PARTICIPATING", "10": "NOT_ALIVE", "100": "NOT_ALIVE", "101": "NON_PARTICIPATING", "102": "NON_PARTICIPATING", "103": "NON_PARTICIPATING", "104": "NOT_ALIVE", "105": "NON_PARTICIPATING", "106": "NON_PARTICIPATING", "107": "NOT_ALIVE", "108": "NON_PARTICIPATING", "109": "NON_PARTICIPATING", "11": "NOT_ALIVE", "110": "NOT_ALIVE", "111": "NON_PARTICIPATING", "112": "SUCCESS", "113": "SUCCESS", "114": "NON_PARTICIPATING", "115": "NON_PARTICIPATING", "116": "NON_PARTICIPATING", "117": "NON_PARTICIPATING", "118": "NON_PARTICIPATING", "119": "NON_PARTICIPATING", "12": "NOT_ALIVE", "120": "NON_PARTICIPATING", "121": "NON_PARTICIPATING", "122": "NON_PARTICIPATING", "123": "NON_PARTICIPATING", "124": "NON_PARTICIPATING", "125": "NOT_ALIVE", "126": "NON_PARTICIPATING", "127": "NOT_ALIVE", "128": "NON_PARTICIPATING", "129": "NON_PARTICIPATING", "13": "NOT_ALIVE", "130": "NON_PARTICIPATING", "131": "NOT_ALIVE", "132": "NON_PARTICIPATING", "133": "SUCCESS", "134": "NON_PARTICIPATING", "135": "NON_PARTICIPATING", "136": "NOT_ALIVE", "137": "NOT_ALIVE", "138": "SUCCESS", "139": "NON_PARTICIPATING", "14": "NOT_ALIVE", "140": "NON_PARTICIPATING", "141": "NOT_ALIVE", "142": "NON_PARTICIPATING", "143": "NON_PARTICIPATING", "144": "NOT_ALIVE", "145": "NOT_ALIVE", "146": "NOT_ALIVE", "147": "NON_PARTICIPATING", "148": "NOT_ALIVE", "149": "NON_PARTICIPATING", "15": "NON_PARTICIPATING", "150": "NOT_ALIVE", "151": "NOT_ALIVE", "152": "NOT_ALIVE", "153": "NON_PARTICIPATING", "154": "NON_PARTICIPATING", "155": "NOT_ALIVE", "156": "NON_PARTICIPATING", "157": "NOT_ALIVE", "158": "NOT_ALIVE", "159": "NOT_ALIVE", "16": "NON_PARTICIPATING", "160": "NOT_ALIVE", "161": "NON_PARTICIPATING", "162": "NON_PARTICIPATING", "163": "NOT_ALIVE", "164": "NON_PARTICIPATING", "165": "NON_PARTICIPATING", "166": "NON_PARTICIPATING", "167": "NON_PARTICIPATING", "168": "NOT_ALIVE", "169": "NOT_ALIVE", "17": "NOT_ALIVE", "170": "NOT_ALIVE", "171": "NOT_ALIVE", "172": "NON_PARTICIPATING", "173": "NON_PARTICIPATING", "174": "NON_PARTICIPATING", "175": "NOT_ALIVE", "176": "NOT_ALIVE", "177": "NON_PARTICIPATING", "178": "NOT_ALIVE", "179": "NOT_ALIVE", "18": "NOT_ALIVE", "180": "NOT_ALIVE", "181": "NOT_ALIVE", "182": "NOT_ALIVE", "183": "NON_PARTICIPATING", "184": "NON_PARTICIPATING", "185": "NOT_ALIVE", "186": "NOT_ALIVE", "187": "NON_PARTICIPATING", "188": "NON_PARTICIPATING", "189": "NOT_ALIVE", "19": "NON_PARTICIPATING", "190": "NOT_ALIVE", "191": "NON_PARTICIPATING", "192": "NOT_ALIVE", "193": "NON_PARTICIPATING", "194": "NON_PARTICIPATING", "195": "NON_PARTICIPATING", "196": "SUCCESS", "197": "NON_PARTICIPATING", "198": "NOT_ALIVE", "199": "NON_PARTICIPATING", "2": "NOT_ALIVE", "20": "NOT_ALIVE", "200": "NOT_ALIVE", "201": "NON_PARTICIPATING", "202": "NON_PARTICIPATING", "203": "NOT_ALIVE", "204": "NOT_ALIVE", "205": "NON_PARTICIPATING", "206": "NON_PARTICIPATING", "207": "NOT_ALIVE", "208": "NON_PARTICIPATING", "209": "NOT_ALIVE", "21": "NOT_ALIVE", "210": "NON_PARTICIPATING", "211": "NON_PARTICIPATING", "212": "SUCCESS", "213": "NON_PARTICIPATING", "214": "NOT_ALIVE", "215": "SUCCESS", "216": "NOT_ALIVE", "217": "NON_PARTICIPATING", "218": "NON_PARTICIPATING", "219": "NOT_ALIVE", "22": "NON_PARTICIPATING", "220": "NON_PARTICIPATING", "221": "NON_PARTICIPATING", "222": "NOT_ALIVE", "223": "NOT_ALIVE", "224": "NOT_ALIVE", "225": "NOT_ALIVE", "226": "NOT_ALIVE", "227": "NOT_ALIVE", "228": "NON_PARTICIPATING", "229": "NON_PARTICIPATING", "23": "NON_PARTICIPATING", "230": "NON_PARTICIPATING", "231": "NOT_ALIVE", "232": "NOT_ALIVE", "233": "NOT_ALIVE", "234": "NOT_ALIVE", "235": "NON_PARTICIPATING", "236": "NON_PARTICIPATING", "237": "NON_PARTICIPATING", "238": "NON_PARTICIPATING", "239": "NOT_ALIVE", "24": "NOT_ALIVE", "240": "NON_PARTICIPATING", "241": "NOT_ALIVE", "242": "NON_PARTICIPATING", "243": "NOT_ALIVE", "244": "NOT_ALIVE", "245": "NON_PARTICIPATING", "246": "NOT_ALIVE", "247": "NON_PARTICIPATING", "248": "NON_PARTICIPATING", "249": "NOT_ALIVE", "25": "NON_PARTICIPATING", "250": "NON_PARTICIPATING", "251": "NOT_ALIVE", "252": "NON_PARTICIPATING", "253": "NOT_ALIVE", "254": "SUCCESS", "255": "NOT_ALIVE", "26": "NON_PARTICIPATING", "27": "NOT_ALIVE", "28": "NOT_ALIVE", "29": "NON_PARTICIPATING", "3": "NON_PARTICIPATING", "30": "SUCCESS", "31": "NON_PARTICIPATING", "32": "NON_PARTICIPATING", "33": "NOT_ALIVE", "34": "NOT_ALIVE", "35": "SUCCESS", "36": "NON_PARTICIPATING", "37": "NOT_ALIVE", "38": "NON_PARTICIPATING", "39": "SUCCESS", "4": "NON_PARTICIPATING", "40": "NON_PARTICIPATING", "41": "NON_PARTICIPATING", "42": "NOT_ALIVE", "43": "NOT_ALIVE", "44": "NON_PARTICIPATING", "45": "NON_PARTICIPATING", "46": "NOT_ALIVE", "47": "SUCCESS", "48": "NON_PARTICIPATING", "49": "NON_PARTICIPATING", "5": "SUCCESS", "50": "NON_PARTICIPATING", "51": "NON_PARTICIPATING", "52": "SUCCESS", "53": "NON_PARTICIPATING", "54": "NOT_ALIVE", "55": "NOT_ALIVE", "56": "NOT_ALIVE", "57": "NON_PARTICIPATING", "58": "NON_PARTICIPATING", "59": "SUCCESS", "6": "NON_PARTICIPATING", "60": "NOT_ALIVE", "61": "NON_PARTICIPATING", "62": "NON_PARTICIPATING", "63": "SUCCESS", "64": "NOT_ALIVE", "65": "NON_PARTICIPATING", "66": "NOT_ALIVE", "67": "NOT_ALIVE", "68": "NOT_ALIVE", "69": "NOT_ALIVE", "7": "NON_PARTICIPATING", "70": "NOT_ALIVE", "71": "NON_PARTICIPATING", "72": "NON_PARTICIPATING", "73": "NON_PARTICIPATING", "74": "NOT_ALIVE", "75": "SUCCESS", "76": "NOT_ALIVE", "77": "SUCCESS", "78": "NOT_ALIVE", "79": "NON_PARTICIPATING", "8": "NOT_ALIVE", "80": "NON_PARTICIPATING", "81": "NON_PARTICIPATING", "82": "NON_PARTICIPATING", "83": "NOT_ALIVE", "84": "NON_PARTICIPATING", "85": "NON_PARTICIPATING", "86": "NON_PARTICIPATING", "87": "NON_PARTICIPATING", "88": "NON_PARTICIPATING", "89": "NOT_ALIVE", "9": "NON_PARTICIPATING", "90": "NOT_ALIVE", "91": "NOT_ALIVE", "92": "SUCCESS", "93": "NOT_ALIVE", "94": "NON_PARTICIPATING", "95": "NOT_ALIVE", "96": "NON_PARTICIPATING", "97": "NON_PARTICIPATING", "98": "NOT_ALIVE", "99": "NOT_ALIVE" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6010342, 6010347, 6010353, 6010358 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 13, "intermediate_size": 5632, "last_allreduce_block": 6009945, "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 }