{ "_name_or_path": "distributed/llama-1b", "all_reduce_scores": { "0": "NOT_ALIVE", "1": "NOT_ALIVE", "10": "NOT_ALIVE", "100": "NOT_ALIVE", "101": "NON_PARTICIPATING", "102": "NON_PARTICIPATING", "103": "SUCCESS", "104": "SUCCESS", "105": "SUCCESS", "106": "SUCCESS", "107": "NOT_ALIVE", "108": "NOT_ALIVE", "109": "SUCCESS", "11": "NOT_ALIVE", "110": "NON_PARTICIPATING", "111": "FAIL", "112": "NON_PARTICIPATING", "113": "NOT_ALIVE", "114": "NOT_ALIVE", "115": "NOT_ALIVE", "116": "SUCCESS", "117": "NOT_ALIVE", "118": "NOT_ALIVE", "119": "NON_PARTICIPATING", "12": "NOT_ALIVE", "120": "SUCCESS", "121": "NON_PARTICIPATING", "122": "SUCCESS", "123": "SUCCESS", "124": "NOT_ALIVE", "125": "SUCCESS", "126": "SUCCESS", "127": "NON_PARTICIPATING", "128": "NON_PARTICIPATING", "129": "FAIL", "13": "NOT_ALIVE", "130": "NON_PARTICIPATING", "131": "NON_PARTICIPATING", "132": "SUCCESS", "133": "SUCCESS", "134": "FAIL", "135": "SUCCESS", "136": "NON_PARTICIPATING", "137": "FAIL", "138": "NOT_ALIVE", "139": "NOT_ALIVE", "14": "NOT_ALIVE", "140": "SUCCESS", "141": "NON_PARTICIPATING", "142": "SUCCESS", "143": "NOT_ALIVE", "144": "NOT_ALIVE", "145": "FAIL", "146": "NOT_ALIVE", "147": "NON_PARTICIPATING", "148": "NON_PARTICIPATING", "149": "SUCCESS", "15": "NOT_ALIVE", "150": "SUCCESS", "151": "NON_PARTICIPATING", "152": "NOT_ALIVE", "153": "SUCCESS", "154": "SUCCESS", "155": "NOT_ALIVE", "156": "NOT_ALIVE", "157": "FAIL", "158": "NON_PARTICIPATING", "159": "SUCCESS", "16": "SUCCESS", "160": "SUCCESS", "161": "NOT_ALIVE", "162": "SUCCESS", "163": "NOT_ALIVE", "164": "SUCCESS", "165": "SUCCESS", "166": "SUCCESS", "167": "FAIL", "168": "SUCCESS", "169": "FAIL", "17": "NOT_ALIVE", "170": "SUCCESS", "171": "NOT_ALIVE", "172": "FAIL", "173": "SUCCESS", "174": "NON_PARTICIPATING", "175": "SUCCESS", "176": "NOT_ALIVE", "177": "SUCCESS", "178": "NON_PARTICIPATING", "179": "SUCCESS", "18": "FAIL", "180": "NOT_ALIVE", "181": "SUCCESS", "182": "NON_PARTICIPATING", "183": "FAIL", "184": "SUCCESS", "185": "NON_PARTICIPATING", "186": "FAIL", "187": "NON_PARTICIPATING", "188": "NOT_ALIVE", "189": "SUCCESS", "19": "NOT_ALIVE", "190": "FAIL", "191": "FAIL", "192": "SUCCESS", "193": "SUCCESS", "194": "SUCCESS", "195": "NON_PARTICIPATING", "196": "NOT_ALIVE", "197": "FAIL", "198": "FAIL", "199": "SUCCESS", "2": "SUCCESS", "20": "NOT_ALIVE", "200": "SUCCESS", "201": "NOT_ALIVE", "202": "NON_PARTICIPATING", "203": "SUCCESS", "204": "SUCCESS", "205": "NOT_ALIVE", "206": "SUCCESS", "207": "NOT_ALIVE", "208": "NON_PARTICIPATING", "209": "NON_PARTICIPATING", "21": "NOT_ALIVE", "210": "NOT_ALIVE", "211": "NON_PARTICIPATING", "212": "NOT_ALIVE", "213": "NOT_ALIVE", "214": "NON_PARTICIPATING", "215": "SUCCESS", "216": "SUCCESS", "217": "SUCCESS", "218": "NON_PARTICIPATING", "219": "NON_PARTICIPATING", "22": "FAIL", "220": "NOT_ALIVE", "221": "NOT_ALIVE", "222": "NON_PARTICIPATING", "223": "NON_PARTICIPATING", "224": "NON_PARTICIPATING", "225": "SUCCESS", "226": "SUCCESS", "227": "NON_PARTICIPATING", "228": "NON_PARTICIPATING", "229": "SUCCESS", "23": "NOT_ALIVE", "230": "NOT_ALIVE", "231": "NON_PARTICIPATING", "232": "NOT_ALIVE", "233": "NOT_ALIVE", "234": "SUCCESS", "235": "NOT_ALIVE", "236": "SUCCESS", "237": "NOT_ALIVE", "238": "NOT_ALIVE", "239": "NOT_ALIVE", "24": "SUCCESS", "240": "SUCCESS", "241": "FAIL", "242": "NON_PARTICIPATING", "243": "SUCCESS", "244": "FAIL", "245": "NOT_ALIVE", "246": "NOT_ALIVE", "247": "SUCCESS", "248": "NOT_ALIVE", "249": "SUCCESS", "25": "SUCCESS", "250": "FAIL", "251": "NON_PARTICIPATING", "252": "NON_PARTICIPATING", "253": "SUCCESS", "254": "NON_PARTICIPATING", "255": "SUCCESS", "26": "SUCCESS", "27": "SUCCESS", "28": "NON_PARTICIPATING", "29": "SUCCESS", "3": "NOT_ALIVE", "30": "NOT_ALIVE", "31": "SUCCESS", "32": "SUCCESS", "33": "NON_PARTICIPATING", "34": "NOT_ALIVE", "35": "SUCCESS", "36": "FAIL", "37": "NOT_ALIVE", "38": "SUCCESS", "39": "NOT_ALIVE", "4": "SUCCESS", "40": "NON_PARTICIPATING", "41": "SUCCESS", "42": "FAIL", "43": "FAIL", "44": "SUCCESS", "45": "NOT_ALIVE", "46": "NOT_ALIVE", "47": "FAIL", "48": "FAIL", "49": "SUCCESS", "5": "NOT_ALIVE", "50": "NOT_ALIVE", "51": "FAIL", "52": "SUCCESS", "53": "FAIL", "54": "NON_PARTICIPATING", "55": "NOT_ALIVE", "56": "NON_PARTICIPATING", "57": "SUCCESS", "58": "NOT_ALIVE", "59": "SUCCESS", "6": "SUCCESS", "60": "SUCCESS", "61": "SUCCESS", "62": "SUCCESS", "63": "NON_PARTICIPATING", "64": "NOT_ALIVE", "65": "NOT_ALIVE", "66": "SUCCESS", "67": "NON_PARTICIPATING", "68": "NON_PARTICIPATING", "69": "NON_PARTICIPATING", "7": "NOT_ALIVE", "70": "NOT_ALIVE", "71": "SUCCESS", "72": "NON_PARTICIPATING", "73": "NON_PARTICIPATING", "74": "SUCCESS", "75": "NOT_ALIVE", "76": "SUCCESS", "77": "SUCCESS", "78": "NON_PARTICIPATING", "79": "NOT_ALIVE", "8": "SUCCESS", "80": "NOT_ALIVE", "81": "NON_PARTICIPATING", "82": "NON_PARTICIPATING", "83": "NOT_ALIVE", "84": "SUCCESS", "85": "SUCCESS", "86": "NON_PARTICIPATING", "87": "FAIL", "88": "NOT_ALIVE", "89": "NOT_ALIVE", "9": "NOT_ALIVE", "90": "NOT_ALIVE", "91": "NON_PARTICIPATING", "92": "NON_PARTICIPATING", "93": "NOT_ALIVE", "94": "SUCCESS", "95": "NON_PARTICIPATING", "96": "FAIL", "97": "FAIL", "98": "NON_PARTICIPATING", "99": "NON_PARTICIPATING" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6406617, 6406634 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 70, "intermediate_size": 5632, "last_allreduce_block": 6406083, "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 }