ini adalah gambaran real bagaimana proses inference model AI bekerja, kebayang gak gimana otak kita bekerja?
ini proses inference model 4b, dengan tambahan block expansion untuk spesialisasi skill coding dan agentic dibangun dari base model qwen 3.5 4b
karena heal model pruning model 9b -> 7b mahal biayanya akhirnya saya coba pindah metode
model 4b -> extend block -> 5b training new block
lanjutkan merge new block ke based model
dan tahap akhir QLoRa 5b r=32 mari kita coba lihat hasilnya nanti semoga bisa lebih bagus dari 4b
Harus susun datasets coding, reasoning dan agentic sebanyak banyaknya dan seberkualitas mungkin
sebelum model model frontier itu terlalu mahal untuk diakses
Tenang, walau 1 dollar skrng udah 18.000
Itu artinya rupiah lagi dilemahin, biar ekspor josss, toh beli tempe goreng masih pake rupiah. Ga pake dolar.
Amanlah.
Itu buktinya ibox, resto mewah, event2 fashion tetep rame.
Ini ulah mata2 asing yg buat gaduh.
Jaya jaya jaya
Sharp volatility in the market right now like what I and @Catto_Man_ wants 😆
Hedging works as expected, will deep dive on the onchain data after I got my morning coffee
Solhedge update, the hedging position looks good, the fees overlap the hedging pnl.
Can't wait to see some volatility in the market so we can learn about how strong this LP-Hedge concept
Trying out @solhedge_prjct for an LP-hedge strategy. Glad the team and I share the same view on LP hedging 😆
Let’s see if this can generate consistent, day-by-day profits 🫡
302 escrows created so far, with 58 users actively running LP + hedge positions.
That activity has already generated $16,365.79 in claimed fees, averaging $54.19 per escrow.
Current TVL: $12,651.53
More users are showing up, staying active, and earning real yield on SolHedge 🛡️