a3 / config.json
zeras141a's picture
Run 8. Outer Step 84. Inner Step 231.
b0b6829 verified
{
"_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": "FAIL",
"103": "NOT_ALIVE",
"104": "NON_PARTICIPATING",
"105": "SUCCESS",
"106": "FAIL",
"107": "NON_PARTICIPATING",
"108": "NON_PARTICIPATING",
"109": "SUCCESS",
"11": "NOT_ALIVE",
"110": "NON_PARTICIPATING",
"111": "NOT_ALIVE",
"112": "NON_PARTICIPATING",
"113": "NOT_ALIVE",
"114": "NON_PARTICIPATING",
"115": "SUCCESS",
"116": "NOT_ALIVE",
"117": "SUCCESS",
"118": "NOT_ALIVE",
"119": "NON_PARTICIPATING",
"12": "NOT_ALIVE",
"120": "SUCCESS",
"121": "NON_PARTICIPATING",
"122": "SUCCESS",
"123": "FAIL",
"124": "NOT_ALIVE",
"125": "NOT_ALIVE",
"126": "FAIL",
"127": "SUCCESS",
"128": "NON_PARTICIPATING",
"129": "NON_PARTICIPATING",
"13": "SUCCESS",
"130": "SUCCESS",
"131": "NON_PARTICIPATING",
"132": "SUCCESS",
"133": "NOT_ALIVE",
"134": "NON_PARTICIPATING",
"135": "SUCCESS",
"136": "NON_PARTICIPATING",
"137": "FAIL",
"138": "NOT_ALIVE",
"139": "SUCCESS",
"14": "NOT_ALIVE",
"140": "SUCCESS",
"141": "NOT_ALIVE",
"142": "SUCCESS",
"143": "SUCCESS",
"144": "NOT_ALIVE",
"145": "SUCCESS",
"146": "NOT_ALIVE",
"147": "NON_PARTICIPATING",
"148": "NON_PARTICIPATING",
"149": "SUCCESS",
"15": "NOT_ALIVE",
"150": "SUCCESS",
"151": "NON_PARTICIPATING",
"152": "NON_PARTICIPATING",
"153": "NOT_ALIVE",
"154": "SUCCESS",
"155": "SUCCESS",
"156": "NOT_ALIVE",
"157": "SUCCESS",
"158": "NON_PARTICIPATING",
"159": "SUCCESS",
"16": "NOT_ALIVE",
"160": "NON_PARTICIPATING",
"161": "SUCCESS",
"162": "SUCCESS",
"163": "SUCCESS",
"164": "NON_PARTICIPATING",
"165": "SUCCESS",
"166": "NOT_ALIVE",
"167": "SUCCESS",
"168": "NON_PARTICIPATING",
"169": "NON_PARTICIPATING",
"17": "SUCCESS",
"170": "NON_PARTICIPATING",
"171": "NOT_ALIVE",
"172": "NOT_ALIVE",
"173": "NON_PARTICIPATING",
"174": "SUCCESS",
"175": "NON_PARTICIPATING",
"176": "NOT_ALIVE",
"177": "SUCCESS",
"178": "FAIL",
"179": "FAIL",
"18": "FAIL",
"180": "SUCCESS",
"181": "NON_PARTICIPATING",
"182": "NON_PARTICIPATING",
"183": "SUCCESS",
"184": "SUCCESS",
"185": "SUCCESS",
"186": "FAIL",
"187": "NON_PARTICIPATING",
"188": "SUCCESS",
"189": "NOT_ALIVE",
"19": "SUCCESS",
"190": "NON_PARTICIPATING",
"191": "FAIL",
"192": "SUCCESS",
"193": "SUCCESS",
"194": "NOT_ALIVE",
"195": "NON_PARTICIPATING",
"196": "SUCCESS",
"197": "NOT_ALIVE",
"198": "FAIL",
"199": "FAIL",
"2": "SUCCESS",
"20": "NOT_ALIVE",
"200": "SUCCESS",
"201": "NOT_ALIVE",
"202": "NON_PARTICIPATING",
"203": "FAIL",
"204": "NON_PARTICIPATING",
"205": "NON_PARTICIPATING",
"206": "SUCCESS",
"207": "NOT_ALIVE",
"208": "SUCCESS",
"209": "NON_PARTICIPATING",
"21": "NOT_ALIVE",
"210": "FAIL",
"211": "SUCCESS",
"212": "SUCCESS",
"213": "NON_PARTICIPATING",
"214": "NOT_ALIVE",
"215": "SUCCESS",
"216": "FAIL",
"217": "SUCCESS",
"218": "SUCCESS",
"219": "SUCCESS",
"22": "FAIL",
"220": "SUCCESS",
"221": "NON_PARTICIPATING",
"222": "SUCCESS",
"223": "NON_PARTICIPATING",
"224": "NON_PARTICIPATING",
"225": "NON_PARTICIPATING",
"226": "NON_PARTICIPATING",
"227": "SUCCESS",
"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": "SUCCESS",
"237": "NON_PARTICIPATING",
"238": "NON_PARTICIPATING",
"239": "NOT_ALIVE",
"24": "NOT_ALIVE",
"240": "FAIL",
"241": "NON_PARTICIPATING",
"242": "NON_PARTICIPATING",
"243": "FAIL",
"244": "SUCCESS",
"245": "SUCCESS",
"246": "SUCCESS",
"247": "FAIL",
"248": "NOT_ALIVE",
"249": "SUCCESS",
"25": "SUCCESS",
"250": "FAIL",
"251": "NOT_ALIVE",
"252": "SUCCESS",
"253": "SUCCESS",
"254": "NON_PARTICIPATING",
"255": "SUCCESS",
"26": "FAIL",
"27": "SUCCESS",
"28": "NON_PARTICIPATING",
"29": "SUCCESS",
"3": "NOT_ALIVE",
"30": "NOT_ALIVE",
"31": "SUCCESS",
"32": "SUCCESS",
"33": "NON_PARTICIPATING",
"34": "NON_PARTICIPATING",
"35": "FAIL",
"36": "FAIL",
"37": "NOT_ALIVE",
"38": "FAIL",
"39": "NOT_ALIVE",
"4": "NOT_ALIVE",
"40": "SUCCESS",
"41": "FAIL",
"42": "SUCCESS",
"43": "SUCCESS",
"44": "FAIL",
"45": "NON_PARTICIPATING",
"46": "NOT_ALIVE",
"47": "NON_PARTICIPATING",
"48": "FAIL",
"49": "NON_PARTICIPATING",
"5": "NOT_ALIVE",
"50": "NOT_ALIVE",
"51": "SUCCESS",
"52": "SUCCESS",
"53": "NOT_ALIVE",
"54": "SUCCESS",
"55": "SUCCESS",
"56": "FAIL",
"57": "SUCCESS",
"58": "SUCCESS",
"59": "NOT_ALIVE",
"6": "NOT_ALIVE",
"60": "NON_PARTICIPATING",
"61": "FAIL",
"62": "NOT_ALIVE",
"63": "NON_PARTICIPATING",
"64": "NON_PARTICIPATING",
"65": "SUCCESS",
"66": "NON_PARTICIPATING",
"67": "NON_PARTICIPATING",
"68": "NON_PARTICIPATING",
"69": "SUCCESS",
"7": "FAIL",
"70": "NOT_ALIVE",
"71": "SUCCESS",
"72": "NON_PARTICIPATING",
"73": "NON_PARTICIPATING",
"74": "SUCCESS",
"75": "NOT_ALIVE",
"76": "SUCCESS",
"77": "NON_PARTICIPATING",
"78": "NON_PARTICIPATING",
"79": "SUCCESS",
"8": "SUCCESS",
"80": "NON_PARTICIPATING",
"81": "NON_PARTICIPATING",
"82": "SUCCESS",
"83": "SUCCESS",
"84": "SUCCESS",
"85": "SUCCESS",
"86": "NON_PARTICIPATING",
"87": "SUCCESS",
"88": "NOT_ALIVE",
"89": "SUCCESS",
"9": "NOT_ALIVE",
"90": "SUCCESS",
"91": "NOT_ALIVE",
"92": "SUCCESS",
"93": "NON_PARTICIPATING",
"94": "SUCCESS",
"95": "FAIL",
"96": "SUCCESS",
"97": "SUCCESS",
"98": "NOT_ALIVE",
"99": "NON_PARTICIPATING"
},
"architectures": [
"LlamaForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"block_list": [
6404137,
6404152
],
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 2048,
"initializer_range": 0.02,
"inner_step": 231,
"intermediate_size": 5632,
"last_allreduce_block": 6402610,
"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
}