{ "_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": "NOT_ALIVE", "105": "NON_PARTICIPATING", "106": "SUCCESS", "107": "SUCCESS", "108": "SUCCESS", "109": "SUCCESS", "11": "SUCCESS", "110": "NON_PARTICIPATING", "111": "NON_PARTICIPATING", "112": "NON_PARTICIPATING", "113": "NOT_ALIVE", "114": "NON_PARTICIPATING", "115": "SUCCESS", "116": "SUCCESS", "117": "SUCCESS", "118": "NON_PARTICIPATING", "119": "NON_PARTICIPATING", "12": "NOT_ALIVE", "120": "FAIL", "121": "NOT_ALIVE", "122": "NON_PARTICIPATING", "123": "SUCCESS", "124": "SUCCESS", "125": "NOT_ALIVE", "126": "FAIL", "127": "SUCCESS", "128": "SUCCESS", "129": "SUCCESS", "13": "SUCCESS", "130": "SUCCESS", "131": "NON_PARTICIPATING", "132": "SUCCESS", "133": "SUCCESS", "134": "SUCCESS", "135": "NON_PARTICIPATING", "136": "NON_PARTICIPATING", "137": "FAIL", "138": "FAIL", "139": "SUCCESS", "14": "NOT_ALIVE", "140": "SUCCESS", "141": "SUCCESS", "142": "SUCCESS", "143": "NON_PARTICIPATING", "144": "NON_PARTICIPATING", "145": "SUCCESS", "146": "NON_PARTICIPATING", "147": "FAIL", "148": "NON_PARTICIPATING", "149": "SUCCESS", "15": "NOT_ALIVE", "150": "SUCCESS", "151": "NON_PARTICIPATING", "152": "NON_PARTICIPATING", "153": "NON_PARTICIPATING", "154": "NON_PARTICIPATING", "155": "SUCCESS", "156": "NOT_ALIVE", "157": "FAIL", "158": "NON_PARTICIPATING", "159": "NON_PARTICIPATING", "16": "SUCCESS", "160": "SUCCESS", "161": "SUCCESS", "162": "NON_PARTICIPATING", "163": "NON_PARTICIPATING", "164": "SUCCESS", "165": "NON_PARTICIPATING", "166": "NON_PARTICIPATING", "167": "SUCCESS", "168": "SUCCESS", "169": "SUCCESS", "17": "NOT_ALIVE", "170": "FAIL", "171": "SUCCESS", "172": "NON_PARTICIPATING", "173": "NOT_ALIVE", "174": "SUCCESS", "175": "SUCCESS", "176": "NOT_ALIVE", "177": "SUCCESS", "178": "NOT_ALIVE", "179": "NOT_ALIVE", "18": "FAIL", "180": "SUCCESS", "181": "NON_PARTICIPATING", "182": "NON_PARTICIPATING", "183": "SUCCESS", "184": "FAIL", "185": "FAIL", "186": "FAIL", "187": "NON_PARTICIPATING", "188": "FAIL", "189": "SUCCESS", "19": "NOT_ALIVE", "190": "NOT_ALIVE", "191": "SUCCESS", "192": "NON_PARTICIPATING", "193": "NON_PARTICIPATING", "194": "NON_PARTICIPATING", "195": "NON_PARTICIPATING", "196": "SUCCESS", "197": "NON_PARTICIPATING", "198": "SUCCESS", "199": "SUCCESS", "2": "FAIL", "20": "SUCCESS", "200": "SUCCESS", "201": "NOT_ALIVE", "202": "SUCCESS", "203": "SUCCESS", "204": "SUCCESS", "205": "SUCCESS", "206": "NOT_ALIVE", "207": "NOT_ALIVE", "208": "NON_PARTICIPATING", "209": "NON_PARTICIPATING", "21": "NOT_ALIVE", "210": "SUCCESS", "211": "NON_PARTICIPATING", "212": "SUCCESS", "213": "NON_PARTICIPATING", "214": "FAIL", "215": "FAIL", "216": "SUCCESS", "217": "SUCCESS", "218": "FAIL", "219": "SUCCESS", "22": "SUCCESS", "220": "SUCCESS", "221": "NON_PARTICIPATING", "222": "NON_PARTICIPATING", "223": "SUCCESS", "224": "NON_PARTICIPATING", "225": "FAIL", "226": "NON_PARTICIPATING", "227": "NON_PARTICIPATING", "228": "NON_PARTICIPATING", "229": "SUCCESS", "23": "SUCCESS", "230": "NON_PARTICIPATING", "231": "NOT_ALIVE", "232": "NOT_ALIVE", "233": "FAIL", "234": "SUCCESS", "235": "NON_PARTICIPATING", "236": "NOT_ALIVE", "237": "NON_PARTICIPATING", "238": "NOT_ALIVE", "239": "NON_PARTICIPATING", "24": "FAIL", "240": "SUCCESS", "241": "NON_PARTICIPATING", "242": "SUCCESS", "243": "NON_PARTICIPATING", "244": "NON_PARTICIPATING", "245": "SUCCESS", "246": "SUCCESS", "247": "NON_PARTICIPATING", "248": "NON_PARTICIPATING", "249": "FAIL", "25": "SUCCESS", "250": "SUCCESS", "251": "NOT_ALIVE", "252": "NON_PARTICIPATING", "253": "SUCCESS", "254": "NON_PARTICIPATING", "255": "NOT_ALIVE", "26": "SUCCESS", "27": "NON_PARTICIPATING", "28": "SUCCESS", "29": "SUCCESS", "3": "NOT_ALIVE", "30": "SUCCESS", "31": "SUCCESS", "32": "SUCCESS", "33": "NON_PARTICIPATING", "34": "SUCCESS", "35": "NOT_ALIVE", "36": "NON_PARTICIPATING", "37": "SUCCESS", "38": "FAIL", "39": "NON_PARTICIPATING", "4": "NOT_ALIVE", "40": "SUCCESS", "41": "SUCCESS", "42": "SUCCESS", "43": "NON_PARTICIPATING", "44": "NON_PARTICIPATING", "45": "NON_PARTICIPATING", "46": "NOT_ALIVE", "47": "NON_PARTICIPATING", "48": "SUCCESS", "49": "NON_PARTICIPATING", "5": "NOT_ALIVE", "50": "NOT_ALIVE", "51": "NON_PARTICIPATING", "52": "FAIL", "53": "SUCCESS", "54": "NON_PARTICIPATING", "55": "SUCCESS", "56": "NON_PARTICIPATING", "57": "SUCCESS", "58": "NOT_ALIVE", "59": "NON_PARTICIPATING", "6": "NOT_ALIVE", "60": "SUCCESS", "61": "NOT_ALIVE", "62": "SUCCESS", "63": "NON_PARTICIPATING", "64": "NON_PARTICIPATING", "65": "SUCCESS", "66": "NON_PARTICIPATING", "67": "NON_PARTICIPATING", "68": "SUCCESS", "69": "SUCCESS", "7": "SUCCESS", "70": "NOT_ALIVE", "71": "SUCCESS", "72": "SUCCESS", "73": "NON_PARTICIPATING", "74": "SUCCESS", "75": "NOT_ALIVE", "76": "SUCCESS", "77": "SUCCESS", "78": "FAIL", "79": "NON_PARTICIPATING", "8": "NOT_ALIVE", "80": "SUCCESS", "81": "NON_PARTICIPATING", "82": "NON_PARTICIPATING", "83": "NOT_ALIVE", "84": "SUCCESS", "85": "SUCCESS", "86": "NON_PARTICIPATING", "87": "SUCCESS", "88": "NOT_ALIVE", "89": "NOT_ALIVE", "9": "NOT_ALIVE", "90": "FAIL", "91": "NON_PARTICIPATING", "92": "FAIL", "93": "NON_PARTICIPATING", "94": "NON_PARTICIPATING", "95": "SUCCESS", "96": "NON_PARTICIPATING", "97": "FAIL", "98": "NON_PARTICIPATING", "99": "NON_PARTICIPATING" }, "architectures": [ "LlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "block_list": [ 6397275, 6397291 ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "inner_step": 18, "intermediate_size": 5632, "last_allreduce_block": 6396547, "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 }