{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "NOT_ALIVE", "10": "NOT_ALIVE", "100": "NON_PARTICIPATING", "101": "NON_PARTICIPATING", "102": "NON_PARTICIPATING", "103": "FAIL", "104": "SUCCESS", "105": "SUCCESS", "106": "FAIL", "107": "NOT_ALIVE", "108": "NON_PARTICIPATING", "109": "SUCCESS", "11": "NOT_ALIVE", "110": "SUCCESS", "111": "NOT_ALIVE", "112": "NOT_ALIVE", "113": "NOT_ALIVE", "114": "NOT_ALIVE", "115": "NOT_ALIVE", "116": "NOT_ALIVE", "117": "NON_PARTICIPATING", "118": "NON_PARTICIPATING", "119": "NON_PARTICIPATING", "12": "NOT_ALIVE", "120": "FAIL", "121": "NON_PARTICIPATING", "122": "SUCCESS", "123": "FAIL", "124": "FAIL", "125": "NOT_ALIVE", "126": "FAIL", "127": "NON_PARTICIPATING", "128": "NON_PARTICIPATING", "129": "NON_PARTICIPATING", "13": "SUCCESS", "130": "SUCCESS", "131": "NON_PARTICIPATING", "132": "SUCCESS", "133": "SUCCESS", "134": "SUCCESS", "135": "SUCCESS", "136": "NON_PARTICIPATING", "137": "FAIL", "138": "NOT_ALIVE", "139": "SUCCESS", "14": "NOT_ALIVE", "140": "NON_PARTICIPATING", "141": "NON_PARTICIPATING", "142": "SUCCESS", "143": "NON_PARTICIPATING", "144": "NOT_ALIVE", "145": "SUCCESS", "146": "NOT_ALIVE", "147": "NON_PARTICIPATING", "148": "SUCCESS", "149": "NON_PARTICIPATING", "15": "NOT_ALIVE", "150": "SUCCESS", "151": "NON_PARTICIPATING", "152": "NON_PARTICIPATING", "153": "NOT_ALIVE", "154": "SUCCESS", "155": "NON_PARTICIPATING", "156": "NOT_ALIVE", "157": "FAIL", "158": "NON_PARTICIPATING", "159": "FAIL", "16": "FAIL", "160": "FAIL", "161": "FAIL", "162": "FAIL", "163": "NON_PARTICIPATING", "164": "SUCCESS", "165": "SUCCESS", "166": "NOT_ALIVE", "167": "FAIL", "168": "NOT_ALIVE", "169": "FAIL", "17": "SUCCESS", "170": "SUCCESS", "171": "NOT_ALIVE", "172": "FAIL", "173": "SUCCESS", "174": "FAIL", "175": "SUCCESS", "176": "NOT_ALIVE", "177": "FAIL", "178": "NON_PARTICIPATING", "179": "SUCCESS", "18": "NOT_ALIVE", "180": "SUCCESS", "181": "NON_PARTICIPATING", "182": "NON_PARTICIPATING", "183": "NOT_ALIVE", "184": "SUCCESS", "185": "FAIL", "186": "SUCCESS", "187": "NON_PARTICIPATING", "188": "SUCCESS", "189": "SUCCESS", "19": "SUCCESS", "190": "FAIL", "191": "NON_PARTICIPATING", "192": "FAIL", "193": "SUCCESS", "194": "NON_PARTICIPATING", "195": "SUCCESS", "196": "NON_PARTICIPATING", "197": "NON_PARTICIPATING", "198": "SUCCESS", "199": "SUCCESS", "2": "SUCCESS", "20": "NOT_ALIVE", "200": "SUCCESS", "201": "NOT_ALIVE", "202": "NON_PARTICIPATING", "203": "SUCCESS", "204": "FAIL", "205": "NON_PARTICIPATING", "206": "SUCCESS", "207": "NOT_ALIVE", "208": "FAIL", "209": "NON_PARTICIPATING", "21": "NOT_ALIVE", "210": "SUCCESS", "211": "NON_PARTICIPATING", "212": "SUCCESS", "213": "NON_PARTICIPATING", "214": "NON_PARTICIPATING", "215": "SUCCESS", "216": "SUCCESS", "217": "SUCCESS", "218": "FAIL", "219": "NON_PARTICIPATING", "22": "FAIL", "220": "NON_PARTICIPATING", "221": "NOT_ALIVE", "222": "SUCCESS", "223": "NON_PARTICIPATING", "224": "NON_PARTICIPATING", "225": "NON_PARTICIPATING", "226": "SUCCESS", "227": "NON_PARTICIPATING", "228": "NON_PARTICIPATING", "229": "SUCCESS", "23": "NOT_ALIVE", "230": "NOT_ALIVE", "231": "NON_PARTICIPATING", "232": "NOT_ALIVE", "233": "NON_PARTICIPATING", "234": "NOT_ALIVE", "235": "NON_PARTICIPATING", "236": "FAIL", "237": "NON_PARTICIPATING", "238": "NOT_ALIVE", "239": "NOT_ALIVE", "24": "NOT_ALIVE", "240": "SUCCESS", "241": "NON_PARTICIPATING", "242": "SUCCESS", "243": "SUCCESS", "244": "NOT_ALIVE", "245": "NON_PARTICIPATING", "246": "FAIL", "247": "FAIL", "248": "NOT_ALIVE", "249": "FAIL", "25": "SUCCESS", "250": "SUCCESS", "251": "NON_PARTICIPATING", "252": "SUCCESS", "253": "SUCCESS", "254": "SUCCESS", "255": "SUCCESS", "26": "SUCCESS", "27": "SUCCESS", "28": "NON_PARTICIPATING", "29": "NON_PARTICIPATING", "3": "NOT_ALIVE", "30": "NOT_ALIVE", "31": "SUCCESS", "32": "SUCCESS", "33": "NON_PARTICIPATING", "34": "NOT_ALIVE", "35": "SUCCESS", "36": "SUCCESS", "37": "NOT_ALIVE", "38": "NON_PARTICIPATING", "39": "NOT_ALIVE", "4": "NOT_ALIVE", "40": "SUCCESS", "41": "NON_PARTICIPATING", "42": "SUCCESS", "43": "SUCCESS", "44": "FAIL", "45": "NOT_ALIVE", "46": "NOT_ALIVE", "47": "SUCCESS", "48": "FAIL", "49": "SUCCESS", "5": "NOT_ALIVE", "50": "NOT_ALIVE", "51": "SUCCESS", "52": "FAIL", "53": "NOT_ALIVE", "54": "NON_PARTICIPATING", "55": "NOT_ALIVE", "56": "SUCCESS", "57": "FAIL", "58": "FAIL", "59": "NON_PARTICIPATING", "6": "NOT_ALIVE", "60": "NON_PARTICIPATING", "61": "SUCCESS", "62": "NOT_ALIVE", "63": "NOT_ALIVE", "64": "NON_PARTICIPATING", "65": "SUCCESS", "66": "NON_PARTICIPATING", "67": "NON_PARTICIPATING", "68": "NON_PARTICIPATING", "69": "FAIL", "7": "SUCCESS", "70": "NOT_ALIVE", "71": "SUCCESS", "72": "NON_PARTICIPATING", "73": "NON_PARTICIPATING", "74": "FAIL", "75": "NOT_ALIVE", "76": "SUCCESS", "77": "SUCCESS", "78": "SUCCESS", "79": "NOT_ALIVE", "8": "NOT_ALIVE", "80": "NON_PARTICIPATING", "81": "NON_PARTICIPATING", "82": "NON_PARTICIPATING", "83": "FAIL", "84": "FAIL", "85": "FAIL", "86": "NON_PARTICIPATING", "87": "NOT_ALIVE", "88": "NOT_ALIVE", "89": "NOT_ALIVE", "9": "NOT_ALIVE", "90": "NON_PARTICIPATING", "91": "NON_PARTICIPATING", "92": "NON_PARTICIPATING", "93": "SUCCESS", "94": "SUCCESS", "95": "FAIL", "96": "SUCCESS", "97": "NOT_ALIVE", "98": "NOT_ALIVE", "99": "NON_PARTICIPATING" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6404754, 6404783 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 9, "intermediate_size": 5632, "last_allreduce_block": 6404715, "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 }