{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "NOT_ALIVE", "10": "NOT_ALIVE", "100": "NON_PARTICIPATING", "101": "SUCCESS", "102": "NON_PARTICIPATING", "103": "NON_PARTICIPATING", "104": "NON_PARTICIPATING", "105": "NON_PARTICIPATING", "106": "FAIL", "107": "SUCCESS", "108": "SUCCESS", "109": "SUCCESS", "11": "SUCCESS", "110": "NON_PARTICIPATING", "111": "FAIL", "112": "NON_PARTICIPATING", "113": "NOT_ALIVE", "114": "NON_PARTICIPATING", "115": "FAIL", "116": "NON_PARTICIPATING", "117": "FAIL", "118": "FAIL", "119": "FAIL", "12": "NOT_ALIVE", "120": "NON_PARTICIPATING", "121": "SUCCESS", "122": "NON_PARTICIPATING", "123": "NOT_ALIVE", "124": "NOT_ALIVE", "125": "NOT_ALIVE", "126": "NON_PARTICIPATING", "127": "SUCCESS", "128": "SUCCESS", "129": "FAIL", "13": "SUCCESS", "130": "SUCCESS", "131": "NON_PARTICIPATING", "132": "FAIL", "133": "SUCCESS", "134": "FAIL", "135": "FAIL", "136": "SUCCESS", "137": "FAIL", "138": "NOT_ALIVE", "139": "SUCCESS", "14": "NOT_ALIVE", "140": "SUCCESS", "141": "NON_PARTICIPATING", "142": "SUCCESS", "143": "FAIL", "144": "NON_PARTICIPATING", "145": "SUCCESS", "146": "NON_PARTICIPATING", "147": "NON_PARTICIPATING", "148": "NOT_ALIVE", "149": "FAIL", "15": "NOT_ALIVE", "150": "NOT_ALIVE", "151": "NON_PARTICIPATING", "152": "NON_PARTICIPATING", "153": "NON_PARTICIPATING", "154": "SUCCESS", "155": "SUCCESS", "156": "NOT_ALIVE", "157": "NON_PARTICIPATING", "158": "NON_PARTICIPATING", "159": "NON_PARTICIPATING", "16": "SUCCESS", "160": "NON_PARTICIPATING", "161": "FAIL", "162": "FAIL", "163": "NON_PARTICIPATING", "164": "FAIL", "165": "SUCCESS", "166": "NOT_ALIVE", "167": "NON_PARTICIPATING", "168": "SUCCESS", "169": "FAIL", "17": "NOT_ALIVE", "170": "SUCCESS", "171": "SUCCESS", "172": "NON_PARTICIPATING", "173": "NOT_ALIVE", "174": "SUCCESS", "175": "SUCCESS", "176": "NOT_ALIVE", "177": "SUCCESS", "178": "NOT_ALIVE", "179": "SUCCESS", "18": "NOT_ALIVE", "180": "SUCCESS", "181": "NON_PARTICIPATING", "182": "SUCCESS", "183": "SUCCESS", "184": "SUCCESS", "185": "SUCCESS", "186": "SUCCESS", "187": "NON_PARTICIPATING", "188": "SUCCESS", "189": "FAIL", "19": "NOT_ALIVE", "190": "NON_PARTICIPATING", "191": "FAIL", "192": "NOT_ALIVE", "193": "NON_PARTICIPATING", "194": "NON_PARTICIPATING", "195": "SUCCESS", "196": "FAIL", "197": "NON_PARTICIPATING", "198": "SUCCESS", "199": "SUCCESS", "2": "SUCCESS", "20": "NOT_ALIVE", "200": "NON_PARTICIPATING", "201": "NOT_ALIVE", "202": "SUCCESS", "203": "FAIL", "204": "FAIL", "205": "SUCCESS", "206": "NON_PARTICIPATING", "207": "NON_PARTICIPATING", "208": "NON_PARTICIPATING", "209": "SUCCESS", "21": "NOT_ALIVE", "210": "NON_PARTICIPATING", "211": "NON_PARTICIPATING", "212": "SUCCESS", "213": "NON_PARTICIPATING", "214": "SUCCESS", "215": "SUCCESS", "216": "FAIL", "217": "SUCCESS", "218": "SUCCESS", "219": "NON_PARTICIPATING", "22": "NOT_ALIVE", "220": "SUCCESS", "221": "NON_PARTICIPATING", "222": "NOT_ALIVE", "223": "SUCCESS", "224": "NON_PARTICIPATING", "225": "SUCCESS", "226": "NOT_ALIVE", "227": "NON_PARTICIPATING", "228": "NON_PARTICIPATING", "229": "FAIL", "23": "NOT_ALIVE", "230": "NON_PARTICIPATING", "231": "NON_PARTICIPATING", "232": "NOT_ALIVE", "233": "NON_PARTICIPATING", "234": "SUCCESS", "235": "NON_PARTICIPATING", "236": "SUCCESS", "237": "NON_PARTICIPATING", "238": "NOT_ALIVE", "239": "NON_PARTICIPATING", "24": "SUCCESS", "240": "NON_PARTICIPATING", "241": "NON_PARTICIPATING", "242": "SUCCESS", "243": "NON_PARTICIPATING", "244": "SUCCESS", "245": "SUCCESS", "246": "SUCCESS", "247": "FAIL", "248": "NON_PARTICIPATING", "249": "NOT_ALIVE", "25": "SUCCESS", "250": "FAIL", "251": "NOT_ALIVE", "252": "SUCCESS", "253": "SUCCESS", "254": "NON_PARTICIPATING", "255": "NOT_ALIVE", "26": "FAIL", "27": "SUCCESS", "28": "SUCCESS", "29": "SUCCESS", "3": "NOT_ALIVE", "30": "NOT_ALIVE", "31": "NON_PARTICIPATING", "32": "SUCCESS", "33": "NOT_ALIVE", "34": "SUCCESS", "35": "SUCCESS", "36": "FAIL", "37": "NON_PARTICIPATING", "38": "NON_PARTICIPATING", "39": "NON_PARTICIPATING", "4": "NOT_ALIVE", "40": "SUCCESS", "41": "NON_PARTICIPATING", "42": "SUCCESS", "43": "NOT_ALIVE", "44": "NON_PARTICIPATING", "45": "NOT_ALIVE", "46": "SUCCESS", "47": "NOT_ALIVE", "48": "SUCCESS", "49": "SUCCESS", "5": "NOT_ALIVE", "50": "NOT_ALIVE", "51": "NON_PARTICIPATING", "52": "NON_PARTICIPATING", "53": "SUCCESS", "54": "NON_PARTICIPATING", "55": "SUCCESS", "56": "NOT_ALIVE", "57": "SUCCESS", "58": "NOT_ALIVE", "59": "NON_PARTICIPATING", "6": "NOT_ALIVE", "60": "SUCCESS", "61": "NON_PARTICIPATING", "62": "SUCCESS", "63": "SUCCESS", "64": "NON_PARTICIPATING", "65": "SUCCESS", "66": "NOT_ALIVE", "67": "NON_PARTICIPATING", "68": "SUCCESS", "69": "SUCCESS", "7": "FAIL", "70": "NOT_ALIVE", "71": "NON_PARTICIPATING", "72": "FAIL", "73": "NON_PARTICIPATING", "74": "FAIL", "75": "NON_PARTICIPATING", "76": "SUCCESS", "77": "SUCCESS", "78": "NON_PARTICIPATING", "79": "SUCCESS", "8": "NOT_ALIVE", "80": "NON_PARTICIPATING", "81": "NOT_ALIVE", "82": "SUCCESS", "83": "FAIL", "84": "SUCCESS", "85": "FAIL", "86": "NOT_ALIVE", "87": "SUCCESS", "88": "NON_PARTICIPATING", "89": "NOT_ALIVE", "9": "NOT_ALIVE", "90": "NON_PARTICIPATING", "91": "NON_PARTICIPATING", "92": "SUCCESS", "93": "FAIL", "94": "NON_PARTICIPATING", "95": "SUCCESS", "96": "FAIL", "97": "SUCCESS", "98": "NON_PARTICIPATING", "99": "NON_PARTICIPATING" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6396352, 6396368, 6396386, 6396401, 6396417 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 72, "intermediate_size": 5632, "last_allreduce_block": 6395861, "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 }