K2 / config.json
Mhair's picture
Run 8. Outer Step 90. Inner Step 36.
978eac2 verified
{
"_name_or_path": "Mhair/K2",
"all_reduce_scores": {
"0": "NOT_ALIVE",
"1": "NOT_ALIVE",
"10": "NOT_ALIVE",
"100": "NON_PARTICIPATING",
"101": "NON_PARTICIPATING",
"102": "NON_PARTICIPATING",
"103": "FAIL",
"104": "FAIL",
"105": "SUCCESS",
"106": "SUCCESS",
"107": "NOT_ALIVE",
"108": "NON_PARTICIPATING",
"109": "FAIL",
"11": "NOT_ALIVE",
"110": "FAIL",
"111": "NOT_ALIVE",
"112": "NON_PARTICIPATING",
"113": "NOT_ALIVE",
"114": "NOT_ALIVE",
"115": "NOT_ALIVE",
"116": "NOT_ALIVE",
"117": "NOT_ALIVE",
"118": "NON_PARTICIPATING",
"119": "NON_PARTICIPATING",
"12": "NOT_ALIVE",
"120": "NON_PARTICIPATING",
"121": "NON_PARTICIPATING",
"122": "SUCCESS",
"123": "SUCCESS",
"124": "NOT_ALIVE",
"125": "FAIL",
"126": "NOT_ALIVE",
"127": "NON_PARTICIPATING",
"128": "NON_PARTICIPATING",
"129": "FAIL",
"13": "FAIL",
"130": "FAIL",
"131": "NON_PARTICIPATING",
"132": "SUCCESS",
"133": "SUCCESS",
"134": "FAIL",
"135": "SUCCESS",
"136": "NON_PARTICIPATING",
"137": "FAIL",
"138": "NOT_ALIVE",
"139": "SUCCESS",
"14": "NOT_ALIVE",
"140": "SUCCESS",
"141": "NON_PARTICIPATING",
"142": "NOT_ALIVE",
"143": "FAIL",
"144": "NOT_ALIVE",
"145": "FAIL",
"146": "NOT_ALIVE",
"147": "NON_PARTICIPATING",
"148": "SUCCESS",
"149": "SUCCESS",
"15": "NOT_ALIVE",
"150": "NON_PARTICIPATING",
"151": "NOT_ALIVE",
"152": "NOT_ALIVE",
"153": "SUCCESS",
"154": "FAIL",
"155": "NON_PARTICIPATING",
"156": "NOT_ALIVE",
"157": "NON_PARTICIPATING",
"158": "NON_PARTICIPATING",
"159": "FAIL",
"16": "SUCCESS",
"160": "SUCCESS",
"161": "SUCCESS",
"162": "FAIL",
"163": "NON_PARTICIPATING",
"164": "NON_PARTICIPATING",
"165": "SUCCESS",
"166": "NOT_ALIVE",
"167": "NOT_ALIVE",
"168": "SUCCESS",
"169": "FAIL",
"17": "FAIL",
"170": "FAIL",
"171": "NOT_ALIVE",
"172": "NOT_ALIVE",
"173": "SUCCESS",
"174": "FAIL",
"175": "SUCCESS",
"176": "NOT_ALIVE",
"177": "SUCCESS",
"178": "NON_PARTICIPATING",
"179": "SUCCESS",
"18": "FAIL",
"180": "SUCCESS",
"181": "FAIL",
"182": "NON_PARTICIPATING",
"183": "SUCCESS",
"184": "SUCCESS",
"185": "NON_PARTICIPATING",
"186": "SUCCESS",
"187": "NON_PARTICIPATING",
"188": "SUCCESS",
"189": "NOT_ALIVE",
"19": "SUCCESS",
"190": "NON_PARTICIPATING",
"191": "NON_PARTICIPATING",
"192": "NON_PARTICIPATING",
"193": "SUCCESS",
"194": "SUCCESS",
"195": "NON_PARTICIPATING",
"196": "NON_PARTICIPATING",
"197": "NON_PARTICIPATING",
"198": "NON_PARTICIPATING",
"199": "NON_PARTICIPATING",
"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": "SUCCESS",
"209": "NON_PARTICIPATING",
"21": "NOT_ALIVE",
"210": "SUCCESS",
"211": "NON_PARTICIPATING",
"212": "SUCCESS",
"213": "NON_PARTICIPATING",
"214": "NOT_ALIVE",
"215": "SUCCESS",
"216": "SUCCESS",
"217": "SUCCESS",
"218": "NON_PARTICIPATING",
"219": "SUCCESS",
"22": "SUCCESS",
"220": "NON_PARTICIPATING",
"221": "NON_PARTICIPATING",
"222": "SUCCESS",
"223": "NOT_ALIVE",
"224": "NOT_ALIVE",
"225": "SUCCESS",
"226": "SUCCESS",
"227": "NON_PARTICIPATING",
"228": "NON_PARTICIPATING",
"229": "NON_PARTICIPATING",
"23": "NOT_ALIVE",
"230": "NOT_ALIVE",
"231": "NON_PARTICIPATING",
"232": "NOT_ALIVE",
"233": "NON_PARTICIPATING",
"234": "FAIL",
"235": "NON_PARTICIPATING",
"236": "SUCCESS",
"237": "NON_PARTICIPATING",
"238": "NOT_ALIVE",
"239": "NOT_ALIVE",
"24": "NOT_ALIVE",
"240": "FAIL",
"241": "FAIL",
"242": "SUCCESS",
"243": "SUCCESS",
"244": "SUCCESS",
"245": "SUCCESS",
"246": "SUCCESS",
"247": "SUCCESS",
"248": "NOT_ALIVE",
"249": "NOT_ALIVE",
"25": "SUCCESS",
"250": "NON_PARTICIPATING",
"251": "SUCCESS",
"252": "SUCCESS",
"253": "SUCCESS",
"254": "SUCCESS",
"255": "SUCCESS",
"26": "NOT_ALIVE",
"27": "SUCCESS",
"28": "NON_PARTICIPATING",
"29": "NON_PARTICIPATING",
"3": "NOT_ALIVE",
"30": "NOT_ALIVE",
"31": "SUCCESS",
"32": "FAIL",
"33": "NON_PARTICIPATING",
"34": "NON_PARTICIPATING",
"35": "FAIL",
"36": "SUCCESS",
"37": "NOT_ALIVE",
"38": "FAIL",
"39": "NOT_ALIVE",
"4": "NOT_ALIVE",
"40": "FAIL",
"41": "FAIL",
"42": "SUCCESS",
"43": "FAIL",
"44": "FAIL",
"45": "NOT_ALIVE",
"46": "NOT_ALIVE",
"47": "FAIL",
"48": "SUCCESS",
"49": "FAIL",
"5": "NOT_ALIVE",
"50": "NOT_ALIVE",
"51": "NOT_ALIVE",
"52": "SUCCESS",
"53": "SUCCESS",
"54": "SUCCESS",
"55": "NOT_ALIVE",
"56": "SUCCESS",
"57": "SUCCESS",
"58": "SUCCESS",
"59": "NON_PARTICIPATING",
"6": "NOT_ALIVE",
"60": "NON_PARTICIPATING",
"61": "SUCCESS",
"62": "SUCCESS",
"63": "NON_PARTICIPATING",
"64": "NON_PARTICIPATING",
"65": "SUCCESS",
"66": "NON_PARTICIPATING",
"67": "NON_PARTICIPATING",
"68": "NON_PARTICIPATING",
"69": "NON_PARTICIPATING",
"7": "SUCCESS",
"70": "NOT_ALIVE",
"71": "SUCCESS",
"72": "NON_PARTICIPATING",
"73": "NON_PARTICIPATING",
"74": "SUCCESS",
"75": "NOT_ALIVE",
"76": "SUCCESS",
"77": "SUCCESS",
"78": "FAIL",
"79": "NON_PARTICIPATING",
"8": "SUCCESS",
"80": "NON_PARTICIPATING",
"81": "NON_PARTICIPATING",
"82": "NON_PARTICIPATING",
"83": "NON_PARTICIPATING",
"84": "FAIL",
"85": "FAIL",
"86": "NON_PARTICIPATING",
"87": "SUCCESS",
"88": "NOT_ALIVE",
"89": "NOT_ALIVE",
"9": "SUCCESS",
"90": "NON_PARTICIPATING",
"91": "NOT_ALIVE",
"92": "NON_PARTICIPATING",
"93": "SUCCESS",
"94": "SUCCESS",
"95": "FAIL",
"96": "SUCCESS",
"97": "NOT_ALIVE",
"98": "NON_PARTICIPATING",
"99": "NON_PARTICIPATING"
},
"architectures": [
"LlamaForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"block_list": [
6409334,
6409400
],
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 2048,
"initializer_range": 0.02,
"inner_step": 36,
"intermediate_size": 5632,
"last_allreduce_block": 6405390,
"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
}